Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Integrated effects of top-down attention and statistical learning during visual search: An EEG study

Integrated effects of top-down attention and statistical learning during visual search: An EEG study The present study aims to investigate how the competition between visual elements is solved by top-down and/or statistical learning (SL) attentional control (AC) mechanisms when active together. We hypothesized that the “winner” element that will undergo further processing is selected either by one AC mechanism that prevails over the other, or by the joint activity of both mechanisms. To test these hypotheses, we conducted a visual search experiment that combined an endogenous cueing protocol (valid vs. neutral cue) and an imbalance of target frequency distribution across locations (high- vs. low-frequency location). The unique and combined effects of top-down control and SL mechanisms were measured on behaviour and amplitudes of three evoked-response potential (ERP) components (i.e., N2pc, P1, CNV) related to attentional processing. Our behavioural results showed better performance for validly cued targets and for targets in the high-frequency location. The two factors were found to interact, so that SL effects emerged only in the absence of top-down guidance. Whereas the CNV and P1 only displayed a main effect of cueing, for the N2pc we observed an interaction between cueing and SL, revealing a cueing effect for targets in the low-frequency condition, but not in the high-frequency condition. Thus, our data support the view that top-down control and SL work in a conjoint, integrated manner during target selection. In particular, SL mechanisms are reduced or even absent when a fully reliable top-down guidance of attention is at play. Keywords N2pc · P1 · Statistical learning · Endogenous cueing · Attention control · Priority map Introduction select the relevant information and tune out what is irrel- evant (Desimone & Duncan, 1995; Reynolds & Chelazzi, In everyday life, the large number of visual inputs com- 2004). This process involves one or multiple attentional con- ing from the environment greatly exceeds our sensory and trol (AC) mechanisms that assign attentional priority to a cognitive processing capacities. Looking for a book in a certain stimulus or location in the visual field. A prominent crowded library can be a difficult task since, at all times, theory of attentional guidance is the priority map theory, all the available visual stimuli compete with each other in which suggests a neural representation of visual space that is order to gain access to further processing. Visual attention topographically organized (Bisley & Goldberg, 2010; Ptak, is the cognitive function that acts as a filter, allowing us to 2012). Depending on the context and time, each location in the visual space is suggested to have a specific level of neuronal activity that is determined by the amount of atten- * Carola Dolci tional priority assigned to that location (Awh et al., 2012; carola.dolci@univr.it Chelazzi et al., 2013; Di Bello et al., 2022; Ipata et al., 2009; Department of Neuroscience, Biomedicine, and Movement Serences & Yantis, 2007). The highest activation peak trig- Science, University of Verona, Strada le Grazie, 8, gers a winner-takes-all process, leading to the target at that 37134 Verona, Italy location being selected (Bisley, 2011; Chelazzi et al., 2014; Department of Experimental Psychology, Ghent University, Macaluso & Doricchi, 2013; Noudoost et al., 2010). Thus, Ghent, Belgium the distribution of attentional resources within the prior- Institut des Sciences Cognitives Marc-Jeannerod, Lyon, ity map would be influenced by the action of different AC France mechanisms. Lyon Neuroscience Research Center, Lyon, France Vol.:(0123456789) 1 3 Attention, Perception, & Psychophysics Individual priority signals may originate from various versus mechanism prevalence in solving the competition sources. Traditionally, they have been separated into two between stimuli. Previous studies are more in line with the main categories: top-down and bottom-up. Top-down (or first hypothesis, arguing that these two mechanisms operate goal-directed) AC is an endogenous process, driven by independently from each other, with the influence of the active volitional selection of items that are relevant to a per- two adding up in a linear manner when engaged at the same son’s goals or instructions (Carrasco, 2011; Egeth & Yantis, time (Duncan & Theeuwes, 2020; Gao & Theeuwes, 2020; 1997; Leber & Egeth, 2006; Parisi et al., 2020; Reynolds Geng & Behrmann, 2005). For instance, Gao and Theeu- & Heeger, 2009). For instance, the competition between wes (2020) showed how SL biased the competition in favour stimuli can be solved by the presence of a central visual cue, of a target that appeared frequently in a ceratian location, which indicates the forthcoming target location and allows compared with a target that appeared in a rare location in the pre-allocation of attentional resources to that position, the array, and that this effect was not affected by top-down facilitating target detection (Posner, 1980). In contrast, bot- attention being directed to one or the other location. At the tom-up attention is an exogenous AC mechanism by which same time, better performance was found when participants attentional resources are automatically allocated toward a could benefit from valid (vs. invalid) information that was salient stimulus with highly noticeable feature properties, given to the top-down AC mechanism, when the target was such as luminance, color, or shape (Theeuwes, 1991, 1994; presented in both the high- and the low-frequency locations Theeuwes & Godijn, 2004; Yantis & Egeth, 1999). (Gao & Theeuwes, 2020). This suggests that both mecha- In recent years, it has been shown that people can implic- nisms can guide independently the attentional selection of itly develop another type of bias specifically linked to the specific spatial locations on the priority map. individual’s previous experience with a given context and/ However, in that study, top-down control was always pre- or stimuli, which can also guide target selection (Awh et al., sent, as participants were instructed to attend to a location in 2012; Ferrante et al., 2018; Jiang, 2018). Therefore, a third the array that could correspond to the location of the upcom- AC category has been introduced: experience-dependent ing target, but not with complete certainty; only on 50% of AC (Awh et al., 2012; Chelazzi & Santandrea, 2018; Fail- the trials the cue indicated the exact target location, whereas ing & Theeuwes, 2018). One of the experience-dependent on the other half of the trials it indicated a location nearby. mechanisms is statistical learning (SL), which allows Thus, it is possible that the cue validity led to the absence of humans, but also other animals, to implicitly extract regu- an observable interaction between the two mechanisms, as larities from the environment even without having explicit the level of uncertainty introduced in the paradigm may have instructions (Aslin & Newport, 2012; Druker & Anderson, prevented a possible interaction between the mechanisms. 2010; Ferrante et al., 2018; Duncan & Theeuwes, 2020; For this reason, in the current study we provided participants for evidence in non-human primates: Newport et al., 2004; with a fully predictive top-down control guidance, by using and chicken: Rosa-Salva et al., 2018; Santolin et al., 2016). a 100% valid cue that pointed to the upcoming target loca- In particular, the probability with which a target element tion and compared it to a condition where top-down control occurs in a specific location was found to induce an atten- was not at a play, where participants were provided with an tional bias toward the location where it is more likely to uninformative neutral cue. appear, without the participant being consciously aware of this (Geng & Behrmann, 2002, 2005). EEG markers of visual selective attention As described above, many studies investigated how visual attention is guided by individual AC mechanisms, but they To further examine the prioritization process in visual do not specify how attentional selection is reconfigured search, in the present study, we investigated the neural when different attentional biases are at a play. Indeed, in mechanism underlying it. Specifically, we focused on three many aspects of everyday life multiple AC mechanisms can well-established evoked-response potential (ERP) compo- act simultaneously, and it is still unclear how they inter- nents related to attentional selection: the cue-related contin- act with one another in the prioritization process and how gent negative variation (CNV), and the target-related P1 and the final attentional choice is established. One possibility is N2pc. When investigating selective attention using visual- that, when acting simultaneously, the activity of all the AC search paradigms, the main EEG marker of interest is the mechanisms is added-up and they jointly contribute to solv- N2pc, which is the negative deflection at posterior electrodes ing the competition in favour of one stimulus. Alternatively, contralateral to the target, typically emerging around 200 one mechanism may prevail over the other, thus exclusively ms from the onset of a lateralized target. Traditionally, the governing target selection. N2pc has been assumed to index the shift of covert attention The aim of the present study was to study the combination toward a task-relevant, or salient, stimulus (Eimer, 1996; of AC mechanisms, specifically, top-down AC and statistical Luck & Hillyard, 1994), but other findings suggest that the learning, and to test the hypotheses of a joint contribution N2pc reflects various aspect of target processing (Kiss et al., 1 3 Attention, Perception, & Psychophysics 2008; Theeuwes, 2010; Zivony et al., 2018). Relevant for visual stimuli appeared on the cued compared with the non- the current study, in a recent work using a very similar task cued side of the array, suggesting that it is an early manifes- to the one used here, Rashal and colleagues (2022; Experi- tation of top-down attentional control (Eimer, 1994; Mangun ment 2) observed an N2pc for targets preceded by a valid & Hillyard, 1991; Van Voorhis & Hillyard, 1977). endogenous cue to the target location, suggesting that the N2pc reflected attentional processes also following topdown Aim and hypotheses of the study deployment of attention to that location. As SL induces a change of attentional priority in favour We devised a visual search task to investigate both isolated of the high-frequency target location, it might be expected and integrated effects of different sources of AC during the that a facilitation of target selection (Ferrante et al., 2018; target selection process. In particular, we focused on top- Geng & Behrmann, 2002)  would be accompanied by a down attention control, which we manipulated via endog- larger N2pc elicited by that target. Still, a recent study enous cueing, and statistical learning, which was manipu- by van Moorselaar and Slagter (2019) found instead a lated by an imbalance of target frequency across locations. reduction of N2pc amplitudes when the target appeared By comparing performance in trials where targets appeared frequently in a certain location. In their study, however, in the high- (HFTL) versus low- frequency target location the target competed with only one other stimulus (i.e., the (LFTL) and were preceded by an informative (valid) or distractor), making the task easier as attentional selection non-informative (neutral) cue, we tested whether top-down was quickly accomplished (see also Rashal et al., 2022, for control and SL, when active together, both contribute to evidence that the N2pc is modulated by difficulty-related assigning attentional priority to a specific spatial location factors). Furthermore, statistical learning in that study was (hypothesis 1) or if one mechanism is blocked by the pres- constrained to target repetitions, with the target appearing ence of the other mechanism (hypothesis 2). Specifically, if at the same location for a number of consecutive trials (4 the two mechanisms operate independently, better perfor- trials) within a sequence, but that location varied across mance and a larger N2pc should be observed for targets in the duration of the experiment. Critically, the modulation the HFTL compared with the LFTL condition irrespective of the N2pc reported by van Moorselaar and Slagter (2019) of the cueing condition. At the same time, cueing effects revealed that the N2pc amplitude, and thus the deployment should emerge regardless of the target location frequency of attentional resources needed for target selection, was condition, and better performance and a larger N2pc should diminished for repeating target location in consecutive tri- emerge following a valid cue when the target appears in both als. That is, the N2pc was larger in the first trial than in the the HFTL and the LFTL (for behavioural evidence, see Gao last trial of the repetition sequence. However, this result & Theeuwes, 2020). Alternatively, if the two mechanisms may be attributed to inter-trial priming and may not apply to interact with each other, we should find that one mechanism a situation where SL is established across the entire experi- affects the other in some way. For example, it is possible ment. As a matter of fact, in classic SL paradigms, target that top-down control blocks the effect of SL, such that its location frequency is associated with just one (or a few) effect can be reduced or even gated by pre-cueing the target spatial location(s) or region(s) across the entire experiment, location, resulting in a smaller difference in performance allowing SL to be reinforced continuously and inducing an and N2pc mean amplitude between targets in the HFTL attentional enhancement in favour of that location. and LFTL following a valid cue compared with the same Two other components related to attentional control are difference in performance and N2pc amplitudes for targets the post-cue CNV and the post-target P1 (Mangun, 1995; following a neutral cue. Similarly, it can be that SL blocks Schevernels et al., 2014; Van Den Berg et al., 2014), the first top-down control. In this case, we should find that the benefit specifically related to top-down control, while the latter is of validly cueing the target location is reduced by the pres- potentially modulated by top-down and bottom-up mecha- ence of target-location frequency imbalance, resulting in a nisms. The CNV is characterized by a slow, negative-going smaller cueing effect on behaviour and N2pc amplitude in waveform normally detected in central areas after the pres- the HFTL compared with the LFTL conditions. entation of a warning stimulus such as a cue (Walter et al., Additionally, we examined two other EEG components 1964), likely reflecting a general preparatory attention dur - mostly related to the top-down control: the P1 during vis- ing the cue-target interval of attentional tasks (e.g., Grent- ual search, and the CNV during the cue-target interval. By ‘t-Jong & Woldorff, 2007). The P1 is the first positive-going looking at the CNV and P1 components, we can investigate ERP component, starting around 90 ms after target-array modulations to top-down attentional orienting pre- and post- onset, and displays increased amplitudes over the occipital stimulus array onset. A larger CNV should emerge follow- scalp contralateral to the precued location (Baumgartner ing a valid compared with a neutral cue, reflecting advance et al., 2018; Mangun & Hillyard, 1991). P1 amplitudes have preparation for selecting the target stimulus (Rashal et al., been demonstrated to be enhanced when the corresponding 2022; Schevernels et al., 2014; Van Den Berg et al., 2014). 1 3 Attention, Perception, & Psychophysics Furthermore, the P1 could also be modulated by the pres- on a given trial, with all stimuli being drawn in the same ence or absence of a valid cue. Specifically, we expected a colour. The choice for two colours, which was not essen- larger P1 following a valid compared with a neutral cue, tial to the present task, and which was fully counterbal- indicating an early stimulus categorization when a stimulus anced across conditions, largely relates to earlier work of is presented in the expected spatial location (Heinze et al., ours using the same global approach (Rashal et al., 2022). 1994; Mangun, 1995). Indeed, Livingstone and colleagues Within each stimulus, there was a small gap (diameter of (2017) demonstrated that P1 indexes an enhanced processing 0.25°) of the same grey colour as the background and posi- for the search item pointed by a valid cue at a stage of vision tioned at the upper or lower part. The target was a bar tilted that precedes attentional selection. ±25° across the vertical axis, whereas the other stimuli that Lastly, even if a modulation of the CNV and P1 have been had to be ignored (distractors) were bars tilted ±25° across most clearly associated with cueing, here we tested if SL was the horizontal axis. Two stimuli were presented in the upper able to affect the general preparation and attentional orient- visual field, two on the horizontal midline and two in the ing pre- and post-stimulus array onset. If so, as for N2pc, we lower visual field (Fig.  1, panel a). Since evidence indicates would expect a larger CNV and P1 for targets in the HFTL, that the N2pc is usually larger in the lower visual field and compared with the LFTL condition, and this effect could on the horizontal meridian than in the upper visual field interact with the cueing manipulation. (Bacigalupo & Luck, 2019; Luck et al., 1997), the target never appeared in the two upper locations, which hence just contained filler items. In each visual search display, Methods six stimuli were presented, centred equidistantly 7° away from a white fixation cross (0.5°×0.5°; RGB: 255, 255, 255; Participants luminance 190.2 cd/m ). Before the onset of the stimulus array, a cue stimulus Twenty-four healthy volunteers (four males; mean age 23.62 was presented around the fixation cross (Fig.  1, panel a). years, SD ±3.4 years) with normal or corrected-to-normal The cue consisted of a geometric shape (dimension: 1.5° visual acuity participated in this experiment. None of them × 1.5°) made up of six separate corners, each pointing at had previously taken part in similar or related studies, and one of the stimulus locations. In the case of the neutral cue, they were naive to the purpose of the present study. At the all the corners were coloured with the same pink (RGB: end of the experiment, they received a fixed monetary com- 120, 0, 90; luminance: 89.5 cd/m ), whereas in the case pensation for their participation (€32.5). All subjects gave of the valid cue, five corners were pink and one was cyan their written informed consent before participation. The blue (RGB: 0, 56, 158; luminance: 81.1 cd/m ), indicat- present study was approved by the ethics committee of the ing in which spatial position the target element would be Faculty of Psychology and Educational Sciences of Ghent presented (Fig. 1, panel a). University (code 2021/09). Experimental design Apparatus and stimuli A central cue presented prior to the target array onset was The experiment was conducted in a dimly lit and quiet room, either valid or neutral. In the valid cue condition, the loca- where participants sat in front of a 24-in. Benq XL2411Z tion of the upcoming target was predicted with a validity of LED monitor controlled by a Dell Optiplex 9020 tower 100%. In the neutral cue condition, the cue did not include with Intel Core i5-4590 processor, at 60-Hz refresh-rate. information about the target location. Importantly, in order The viewing distance was held constant at 60 cm by using to not mix the two AC manipulations, SL was manipulated an adjustable chin rest. The experiment was run with the exclusively following neutral cues by introducing, unbe- PsychoPy (v1.84.2) software (Peirce, 2007). Good central known to the participants, an imbalance of target frequency fixation by the participants was monitored using the cam - appearance across the four possible target locations: high, era of an Eyelink 1000 plus (SR Research, Canada). The low and two intermediate location frequencies (Ferrante experimenter was sitting in a different room and warned et al., 2018). The valid cue, when present, indicated with the participants during breaks in case eye-movements were equal frequency each of the four possible target locations (96 observed in the preceding block, to allow correction. trials each location). The neutral cue trials (1,216 trials; 76% The stimuli were rectangular bars of 2.0° × 0.5° in size, of all trials) provided a baseline where the SL effect could be either green (RGB coordinates: 0, 86, 0; luminance: 138.5 assessed in the absence of top-down guidance. Here the tar- cd/m ) or red (RGB values: 170, 0, 0; luminance: 64.8 cd/ get appeared in the high-frequency location 50% of the trials m ), presented on a grey background (RGB: 128, 128, 128; (608 trials), in the low-frequency location for 7.9% of the luminance: 85.5 cd/m ). Colours were randomly chosen trials (96 trials), and in each of the intermediate-frequency 1 3 Attention, Perception, & Psychophysics Fig. 1 a Examples of the trial sequence. Top row: a neutral cue pre- and was the bar tilted ±25° from its vertical axis, while the non-tar- ceded target array onset. Bottom row: target location was predicted by gets were tilted ±25° from their horizontal axis. b Target frequency a valid cue. The target is indicated in the figure by the dashed circle distribution across groups (during neutral cue trials only). Note that (for illustration purposes; no such circle was present during the task) the target never appeared in the two upper locations locations for 21% of these trials (256 trials each). We did duration of the experiment. After a random interval jittered not introduce an imbalance of target frequency appearance between 100 and 300 ms, the cue appeared for 480 ms. After in the valid cue condition because doing so would mean a cue-target interval (CTI), jittered between 700 and 900 ms, introducing an imbalance of valid cues. This, in turn, would the search display appeared and remained visible for 300 complicate the interpretation of the results, as it would be ms. Responses were recorded from the onset of the search impossible to disentangle the benefit in target detection due display until 1,200 ms after display offset, for a total of 1,500 to SL of the target location, or SL of the valid cue, or both. ms. Afterwards, a new trial sequence started automatically. Participants were randomly assigned to one of four groups The task was to discriminate the position of the gap within (Fig. 1, panel b), each with a different spatial configuration the target item (top vs. bottom) by pressing the letter ‘M’ on of target-location probabilities, but with the constraint that the keyboard with their right index finger if the gap was in the high-probability and low-probability conditions were the lower part, or the letter ‘Z’ with their left index finger always in opposite locations in the left versus right visual if it was in the upper part. The experiment included a total field. of 1,600 trials, divided into eight blocks. Before starting the actual experiment, a practice phase of 64 trials was used to Procedure allow participants to familiarise themselves with the task. All the conditions previously described were presented in a Each experimental trial (Fig. 1, panel a) started with a fix- fully randomized order. Participants were instructed to main- ation cross, which remained on the screen for the whole tain their eyes on the fixation cross, and fixation quality was 1 3 Attention, Perception, & Psychophysics monitored by the experimenter by means of the online eye- Liesefeld et al., 2017; Rashal et al., 2022). To determine position display of the eye-tracker. the analysis time-windows for each of these EEG markers, In order to evaluate if participants were aware of the fre- we took the canonical values used in the literature: for the quency manipulation, a survey was conducted at the end of CNV, we selected a time-range from 700 ms after cue onset the experiment (see Ferrante et al., 2018). Participants were until approximately the earliest point in the CTI in which first asked to report whether they noticed something about the search display could appear (plus 70 ms, accounting the spatial distribution of target stimuli, and in case they for transduction delay into visual cortex), i.e., 700–1,250 responded “yes”, they had to report (or guess) the locations ms (e.g., Liebrand et al., 2017; Rashal et al., 2022). For where the target was presented most frequently. the N2pc and P1, the respective time-ranges were set to 200–300 ms (N2pc) and 90–140 ms (P1) after the search- Electrophysiological recording and analysis display onset, in line with the existing literature (e.g., Eimer, 1996; Luck et al., 2000; Mangun & Hillyard, 1991). Note EEG data were recorded using a Brain Products actiCHamp that counter to most of the earlier N2pc and P1 literature, the 64-channel system (Brain Products, Gilching, Germany) target location frequency imbalance led to the fact that for a with 64 active scalp electrodes positioned according to the given participant, the contralateral and ipsilateral locations standard international 10–10 system. Signals were recorded were either PO7 or PO8, and could not be collapsed across at a 500-Hz sampling rate, using Fz as the online reference those locations for different conditions, with corresponding and then re-referenced offline to the average of TP9 and targets on the left and right (e.g., Wu et al., 2011). Therefore, TP10, corresponding to the left and right mastoids. Fz was in this study, the average across locations was possible only then restored to the dataset. A high-pass filter of 0.1 Hz across groups (Wang et al., 2019). was applied to the raw data and segments of the continuous Analyses were performed using R 3.6.2 (R Core Team, data, with clearly identifiable, large artefacts (not including 2016) with ez (Lawrence, 2011/2015) and effectsize blinks and eye movements) were excluded by manual inspec- (Ben-Shachar et al., 2020) packages. For CNV we used tion. Successively, independent component analysis (ICA) rm-ANOVAs to compare the mean amplitude in the dif- was used to remove components related to eye blinks and ferent conditions, whereas for N2pc and P1 we first cal- (residual) eye movements. We then segmented the data into culated the mean amplitude of the ipsi and contra location epochs from −200 ms to 2,900 ms relative to cue onset and of interest, and then performed rm-ANOVAs to compare from −200 ms to 800 ms relative to the stimuli array onset. the difference waves (DWs) resulting from the subtrac- We then baseline-corrected with regard to the 200-ms pre- tion contra-minus-ipsi between different conditions. All cue or pre-stimuli period, respectively. Then, a second arti- these analyses were performed only using trials with cor- fact rejection (AR) was performed to flag and remove epochs rect responses. P values were corrected with Greenhouse- that contained artefacts in the analysed channels (PO7/8; Geisser epsilon in cases of significant sphericity violation. absolute amplitude exceeding ±100 μV). On average this led to exclusion less than 3% of the total trials. In order to study the temporal dynamics of attentional Results orienting and subsequent visual search, we focused on three components, namely the cue-evoked CNV and the P1 and Behaviour N2pc elicited by the search array. The CNV was examined at Cz using the cue-locked epochs (Verleger et al., 1999; In order to assess the effects of and interaction between sta- Rashal et al., 2022), whereas for the N2pc and P1 we used tistical learning and top-down mechanisms, 2 × 2 rm-ANO- the average of two electrodes capturing activity at PO7/ VAs were conducted with Target Location Frequency (high, PO8, where the N2pc and P1 are usually the largest (e.g., low) and Cue (valid, neutral) for accuracy and reaction time (RT). These analyses showed significant main effects of Cue for accuracy and RT [ACC: F(1, 23) = 15.32, p = 0.0006, 2 2 η = 0.39; RT: F(1, 23) = 134.41, p < 0.0001, η = 0.85], p p In order to maintain the same experimental protocol as in our previ- and Target Location Frequency for RT [ACC: F(1, 23) = ous related work, the cap was placed slightly further to the back than 2 2 0.46, p = 0.50, η = 0.01; RT: F(1, 23) = 6.10, p = 0.02, η p p typical, positioning FCz at the Cz site (Rashal et al., 2022). Thus, to = 0.20]. Importantly, a significant interaction between the test the N2pc and P1 components we report the results from the aver- two factors was observed for RT [ACC: F(1, 23) = 0.71, p = age between P3/P5 and P4/P6, which are the channel pairs that were 2 2 closest (~1 cm off) to the PO7/PO8 sites. Similarly, for the CNV the 0.40, η = 0.03; RT: F(1, 23) = 9.32, p = 0.005, η = 0.29]. p p channel used was FCz, which directly corresponds to Cz on the scalp Post hoc paired t-tests (two-tailed) revealed that participants (Verleger et al., 1999; Rashal et al., 2022). For simplicity, we refer to were significantly faster in detecting the target in the HFTL the actual scalp locations, rather than the labels on the cap. compared with the LFTL, but only when the cue was neutral 1 3 Attention, Perception, & Psychophysics Fig. 2 Mean accuracies (left) and reaction times (RTs; right) as a function of cue and target frequency conditions. Error bars represent the standard error of the mean [t(23) = -2.84; p = 0.009, Cohen’s d = -0.34, -30 ms], and the analysis did not change the main results, corroborating not when the cue was valid [t(23) = -0.73; p = 0.47, Cohen’s the implicit nature of the learning process. d = -0.04, -3 ms], suggesting that top-down control is able to exert a gating effect over SL mechanisms. Furthermore, N2pc the benefit of the valid cue was observed in both the HFTL [t(23) = 12.25; p < 0.0001, Cohen’s d = 1.02, 100 ms] and To investigate the effect of top-down AC and SL on the the LFTL [t(23) = 9.75; p < 0.0001, Cohen’s d = 1.49, 130 N2pc, we first investigated if the component was present ms] conditions (Fig. 2), but with a larger benefit for the cue in each condition. One-sample t-tests (one-tailed) showed in the latter (130 vs. 100 ms). a marginally significant N2pc (mean amplitude lesser than At the end of the experimental session, four participants zero) for targets at the HFTL (following neutral cue: t(23) reported having noticed something peculiar regarding the = −1.59, p = 0.06, Cohen's d = −0.32; following valid cue: target frequency, and identified the correct high-frequency t(23) = −1.57, p = 0.06, Cohen's d = −0.32), but not for spatial location as the location where the target was more targets at the LFTL (following neutral cue: t(23) = 0.25, p = likely to appear. However, excluding these participants from 0.59, Cohen's d = 0.05; following valid cue: t(23) = −0.90, p = 0.18, Cohen's d = −0.18). An rm-ANOVA was then conducted with Cue (valid, neutral) and Target Location Frequency (high, low). This analysis considered the contra-minus-ipsi difference waves, directly focusing on attentional lateralization effects. This To examine whether the SL effect was location-specific or, instead, analysis showed a significant main effect of Cue [F (1,23) if it was linked to a hemifield-based representation of space, we 2 = 5.84, p = 0.023, η = 0.20], but not of Target Location repeated the same analysis on the two intermediate locations that were associated with the hemisphere that contained the HTLF and LTLF of each participant. We performed a 2 × 2 rm-ANOVA with Cue (valid, neutral) and Target Location Frequency (intermediate- A 2 × 2 rm-ANOVA with Cue (valid, neutral) and Target Location high, intermediate-low). For accuracy, this analysis showed a signifi- Frequency (high, low) was performed excluding the four subjects who cant main effect of Cue [F(1, 23) = 10.43, p = 0.003, η = 0.31], reported having explicitly noticed the target frequency imbalance. but not of Target Location Frequency [F(1, 23) = 0.16, p = 0.69, For RTs, the analysis confirmed the two main effects [Cue: F(1,19) η = 0.006], and no significant interaction between the two factors = 115.04, p < 0.001, η = 0.85; Target Location Frequency: F(1,19) [F(1, 23) = 1.68, p = 0.20, η = 0.06]. However, for RT, significant = 10.92, p = 0.003, η = 0.36] and a significant interaction [F(1,19) main effects were found for Cue [F(1, 23) = 97.04, p < 0.0001, η = 10.15, p = 0.004, η = 0.34]. A post hoc t-test also confirmed the = 0.80] and Target Location Frequency [F(1, 23) = 8.27, p = 0.008, prevalence of top-down AC over SL, which could be seen only in η = 0.26], but the interaction between these factors was not signifi- the neutral cue trials (neutral: [t(19) = -3.41, p = 0.002, Cohen’s d = cant [F(1, 23) = 2.21, p = 0.15, η = 0.08]. In contrast to the main -0.43, -40 sec]; valid: [t(19) = -1.57, p = 0.13, Cohen’s d = -0.08, -7 analysis, the effect of Target Location Frequency was in the opposite ms]). However, for accuracy we only found a main effect of cue [Cue: direction, showing shorter RTs in the intermediate-low (vs. interme- F(1,19) = 10.63, p = 0.004, η = 0.35; Target Location Frequency: diate-high) condition. Since in the present study we wanted to assess F(1,19) = 1.23, p = 0.28, η = 0.06], and no significant interaction the top-down control in the presence of a clear SL effect, we did not [F(1,19) = 0.41, p = 0.52, η = 0.02]. consider the intermediate frequency locations further. 1 3 Attention, Perception, & Psychophysics 1 3 Attention, Perception, & Psychophysics ◂Fig. 3 Sensor plots showing contra (black line), ipsi (red line) and cue (t(23) = 0.76, p = 0.22, Cohen's d = 0.15), and not for the difference waves (contra-minus-ipsi; blue line) activity following targets in HFTL (following neutral cue: t(23) = −0.53, p = a neutral cue (a, b), or a valid cue (c, d). Panels a and c depict activ- 0.70, Cohen's d = −0.10; following valid cue: t(23) = 0.55, ity in the LFTL condition, and panels b and d depict activity in the p = 0.29, Cohen's d = 0.11). HFTL condition. Time-point zero indicates the search-display onset. The grey area is the time-window where mean amplitude of the P1 A two-way rm-ANOVA with Cue (valid, neutral) and Tar- was calculated, whereas the yellow area refers to the N2pc time- get Location Frequency (high, low) was performed to inves- range. Panels e and f show the mean amplitude of P1 (e) and N2pc tigate the effects of top-down AC and SL on the early stage (f), calculated by subtracting the contra-minus-ipsi channel, in the of target selection. Importantly, this was done on the contra- two Target Location Frequency conditions as a function of the cue. Error bars in plots e and f represent the standard errors of the mean minus-ipsi difference waves, hence characterizing lateraliza- tion effects. Similar to the CNV, this analysis revealed a sig- nificant main effect of Cue [F (1,23) = 23.20, p < 0.001, η Frequency [F(1,23) = 0.47, p = 0.497, η = 0.02]. Impor- = 0.50] that elicited a larger P1 lateralization for valid (vs. tantly, a significant interaction emerged between the two neutral) cues, but not of Target Location Frequency [F(1,23) 2 2 factors [F(1,23) = 4.28, p = 0.049, η = 0.15]. Post hoc = 0.65, p = 0.424, η = 0.02]. Furthermore, no significant p p paired t-tests (two-tailed) revealed that targets in the LFTL interaction emerged between the two factors [F(1,23) = 0.58, condition following a valid cue elicited a larger N2pc com- p = 0.451, η = 0.02] (Fig. 3). pared with targets at that location following a neutral cue [t(23) = 2.85, p = 0.009, Cohen’s d = 0.22; 0.74 μV]. In contrast, this effect was not present for targets in the HFTL Discussion condition [t(23) = -0.10, p = 0.918, Cohen’s d = -0.006; -0.02 μV]. Furthermore, no significant difference in N2pc In the current study, we aimed to assess the combined amplitudes was found between HFTL and LFTL, either in effects and neural correlates of top-down AC and SL, the neutral [t(23) = -0.94, p = 0.355, Cohen’s d = -0.37; when both are present. To that end, we manipulated top- -1.28 μV] or in the valid cue condition [t(23) = -0.40, p = down AC via endogenous cueing, and we introduced an 0.688, Cohen’s d = -0.16; -0.51 μV] (Fig. 3). imbalance of in-target frequency across locations in the same visual search task. Critically, we implemented the CNV target location imbalance only for neutrally cued trials in order to fully dissociate target location frequency To assess whether a valid endogenous cue elicited a pre- and cue validity in our task. Furthermore, we utilized a paratory effect, a one-way rm-ANOVA was performed neutral rather than an invalid cue as a baseline. Impor- with Cue (valid, neutral) on the CNV component. This tantly, to be able to compare and unify results regarding analysis showed a significant difference between the two the interaction between different AC mechanisms, and conditions [F(1,23) = 33.46, p < 0.001, η = 0.59]. Spe- to shed light on the functional architecture of visual cifically, a larger CNV was evoked by valid cues, indicating spatial attention, we used the same visual search task that the participants could prepare to orient their attentional (with some adjustments due to methodological reasons) resources before the search array onset following an inform- already implemented and adapted for the study of the ative cue. Furthermore, we explored whether SL proactively integrated effect of other AC signals, namely top-down modulates top-down control, such that a preparatory effect control via endogenous cueing and bottom-up allocation would emerge according to the target location frequencies. of attention due to salience (Beffara et al., 2022; Rashal To this end, another rm-ANOVA was conducted on the data et al., 2022). from trials following a valid cue in the HFTL and LFTL conditions. No significant difference was found between Combined effect of top‑down control and SL these two conditions [F(1,23) = 0.70, p = 0.41, η = 0.02] on behaviour and on N2pc (Fig. 4). The behavioural results concerning SL and top-down con- P1 trol confirmed our hypotheses and were in line with the literature, showing an overall effect of both mechanisms Similar to the analysis conducted for the N2pc, we per- of facilitation of target identification following valid, com- formed a one-sample t-test (one-tailed) to test whether P1 pared to neutral, cues (e.g., Folk et al., 1992; Posner, 1980; was meaningfully lateralized in each condition (mean ampli- Rashal et al., 2022), as well as targets presented in the high- tude greater than zero). Results showed a significant laterali- (vs. low-) frequency location (Ferrante et al., 2018; Geng zation for targets in the LFTL following a valid cue (t(23) = & Behrmann, 2005). Participants could indeed benefit from 2.30, p = 0.01, Cohen's d = 0.46), but not following a neutral the available information and prepare for the onset of the 1 3 Attention, Perception, & Psychophysics Fig. 4 The plot on the left shows the CNV (contingent negative varia- ing to the LFTL (dashed line) and HFTL (dotted line). Time-point tion) elicited by neutral (black line) and valid (red line) cues, whereas zero indicates cue onset. The yellow area represents the time-window the plot on the right shows the CNV elicited by the valid cue point- where the mean amplitude of the CNV was quantified array, and then efficiently identify the relevant item (i.e., tar - a gating mechanism in order to fully and efficiently guide get). Furthermore, participants’ performance indicated that attention to current objectives. they had learnt the bias induced by the statistical imbalance Previous studies, however, are more in line with the idea of target frequency across locations, which could facilitate that top-down control and SL are independent mechanisms target detection in the location where it was more likely and, when active together, the effects of the two are summed- to appear. During the debriefing at the end of the experi- up to bias the competition over attentional resources (Gao mental session, only four subjects reported having noticed & Theeuwes, 2020; Geng & Behrmann, 2005). Gao and the manipulation. The main results did not significantly Theeuwes (2020) argued that at the level of neural activ- change by excluding their data, supporting the idea that ity, statistical learning creates an implicit landscape where people can implicitly extract regularities from their exter- multiple spatial locations have a certain level of activations nal environment even without explicit instructions (Ferrante and inhibitions, and top-down control may then operate to et al., 2018; Fiser & Aslin, 2001; Jiang, 2018; Saffran et al., orient the attentional spotlight from one location to another 1996). (Gao & Theeuwes, 2020). Their theory seems to be in line Most importantly, in this study we directly examined with our EEG results. Indeed, even if in our study the final whether these two AC signals affect attentional deploy - attentional choice seems to be guided by top-down control ment in an independent manner when acting together, or that prevails over SL in determining performance (e.g., RTs), whether the effect of one signal interferes with the effect the EEG data showed that SL is not completely overridden of the other. Our results were more in line with the lat- by top-down AC, as an interaction is demonstrated by the ter hypothesis, revealing an interaction between the two N2pc modulation. sources of AC. In particular, at the behavioural level, we That is, in terms of the N2pc, the benefit of cueing the observed a gating effect of top-down control over SL, as target location was associated with a larger N2pc that was the effect of SL clearly emerged only in the absence of elicited by targets following a valid, compared to a neutral, top-down guidance; responses were faster for targets in the cue. In addition, this effect interacted with SL, showing an HFTL compared with the LFTL following a neutral but not increased N2pc amplitude for validly cued targets (com- a valid cue. The prevalence of top-down control could rep- pared with neutrally cued targets), but only when shown resent an important feature of the functional architecture at a low-frequency location. This appears to be in line with of visual spatial attention, and it could reflect the ability the priority map theory, whereby the low-frequency target of voluntary control to actively inhibit (or at least reduce) location in general should be associated with less neuronal the contribution of other signals, for example, SL, through activation due to the statistical learning mechanism, and 1 3 Attention, Perception, & Psychophysics thus can benefit from the allocation of top-down attentional across locations, while keeping a location in spatial work- resources, as guided by a valid cue. In contrast, when the ing memory in order to induce top-down control (Awh & target appeared at the high-frequency location, any potential Jonides, 2001; Munneke et al., 2010). Therefore, in their benefit from the preceding cue was abolished. This might experiment, only SL impacted the search strategy, while suggest that the gating effect exerted by top-down AC over the cue determined the second response. Furthermore, in SL that we observed on behaviour was not a full gating; on the present study, there was no invalid cue condition and the contrary, at least at some stage along the target selec- the informative cue predicted the target location with 100% tion process the SL was still exerting an effect that could be validity. Thus, participants could fully trust the informa- strong enough to prevent top-down control from emerging tion coming from the top-down control mechanism. There- (as observed for targets at the high-frequency location in the fore, when these AC mechanisms act together, the inter- current study). action between SL and top-down control may depend on Interestingly, the different pattern of N2pc amplitudes the degree of validity of the latter, in turn determining the observed in the two location-frequency conditions could strength with which it can guide attentional selection even suggest that a serial search strategy was employed by the to the point of bypassing the information coming from SL. participants to find the target (e.g., Woodman & Luck, 2003). Together, our findings suggest a close and complex inter - That is, the participants may have first searched the high-fre- action between top-down control and SL, where when one quency location before moving to the intermediate- and low- mechanism is acting, and is potentially strong enough to frequency locations. The positive peak observed in the low- optimize selection, the effect of the other is reduced. frequency condition at the N2pc time-window supports this However, we have to note that although the present view (Fig. 3a), considering that the high- and low-frequency experiment contained a number of trials that should be suf- locations were positioned at exactly opposite from each other ficient to detect significant effects on the EEG markers of on the screen. Still, this positive peak does not seem to be a interest with our sample size (Boudewyn et al., 2018), it fully inverted N2pc, thus, at this point this is only a specula- seems to be limited in its power to capture smaller effects tion that could benefit from further investigation. we observe as significant (e.g. Ngiam et al., 2021). Given One could argue that the size of the N2pc was reduced that a specific target-frequency condition was associated due to the target appearing at the same location repeatedly with just one hemisphere depending on the subject group across consecutive trials, as there was less need for the allo- (e.g., HFTL in the right hemisphere), in the current study cation of attentional resources if attention was already at the the N2pc was calculated by collapsing the left and right correct location. This seems consistent with the cueing effect hemispheres only at a group level (see Wang et al., 2019, for we observed, since the SL manipulation was present only in a similar experimental paradigm). This might have made the the neutrally cued trials. That is, the target was more likely data vulnerable to increased noise from residual inter-indi- to appear repeatedly at the same location in the neutral cue vidual variability related to asymmetries in brain activity. condition than in the valid cue condition. In contrast, in the This, in turn, might have caused reduced statistical power in valid cue condition there was less chance of repeating target some conditions. Therefore, in future studies, larger sample location across trials, as target location was equally probable sizes and number of trials might be adopted to increase the in that condition. In this vein, the decrease in mean N2pc robustness of SL and cueing effects studied here. amplitude in the neutral cue trials is in line with the find- ings of van Moorselaar and Slagter (2019), who showed a Neural activity underlying the preparatory effect reduction in N2pc amplitude for targets presented at the end of top‑down control of a sequence of targets presented at one location, compared to the targets presented at the beginning of the sequence. In this study, the CNV results reflected processes involved However, this alternative explanation cannot account for in the preparation of anticipatory attention for the upcoming the SL effect we observed, as it predicts a larger N2pc in stimulus and motor preparation needed to respond (Brunia & the LFTL compared to the HFLT, the latter having more van Boxtel, 2001; Tecce, 1972). It has been shown that the chances of repeating target location in consecutive trials than CNV mean amplitude was modulated by the presentation of the former. Evidently, this is the opposite of the pattern we a warning stimulus, such as a cue, demonstrating its link to observed here. strong attentional engagement (e.g., Rashal et al., 2022; Sch- One of the main differences between this study and the evernels et al., 2014; van den Berg et al., 2014). During the earlier ones (e.g., Gao & Theeuwes, 2020) is that in our CTI, a larger CNV was indeed elicited by a valid (vs. neu- paradigm both top-down control and SL directly determine tral) cue, in accordance with the idea that under top-down the search strategy, making it possible to investigate how control participants could orient their attention in advance much SL leaks onto cued trials. In contrast, Gao and Theeu- toward a certain position. In addition, as a consequence of wes (2020) manipulated the frequency of target occurrence the preparatory effect exerted by top-down control, we found 1 3 Attention, Perception, & Psychophysics that the valid cue could also affect the early components of general gating effect of top-down control, but might simply target selection, such as the P1 (e.g., Mangun & Hillyard, index that the SL became a condition-dependent mechanism, 1991). Knowing the upcoming target location allows an in this case, following the neutral cue. However, this hypoth- early allocation of attentional resources toward a specific esis can be excluded given the presence of an interaction region of that display. Indeed, targets following a valid cue between top-down control and SL on the N2pc, which sug- produced a general contralateral enhancement of P1 ampli- gests an intervention of SL in assigning different attentional tudes. Importantly, we did not find any preparatory advan- priorities to different locations on the priority map, which tage due to statistical regularities for the two components; can lead to a reduction of the benefit of top-down control no difference emerged between the CNV elicited by valid during the early stage of target selection. cues pointing to the high-frequency location and valid cues One critical aspect is that the ability of top-down con- pointing to the low-frequency location, and the P1 to the trol to modulate attention and bypass the information of target was not modulated by the target location. all the other AC mechanisms can depend on its strength and its relevance in the given experimental context. The priority map is modulated by combined effects Indeed, a fully reliable informative cue, as the one used in of different attention control signals this experiment, can strongly guide attention toward the instructed location without additional information from Most studies of attention control, however, use experimental other AC signals. Contrary to that, when the informative paradigms that address only one specific AC mechanism at a cue is partially predictive (such as is the case where some time, which makes it difficult to understand the contribution invalid cue trials occurred), the gating effect of top-down of each attentional signal in generating the final attentional control is weaker since it is needs to also consider infor- choice, and controls how goal-directed behaviour is accom- mation coming from other AC signals, in this case, SL. plished by the brain. Here we used a visual search paradigm previously imple- mented in its general form in another study that examined the combined effect of different attention-control sources Conclusion with behaviour and EEG, in an attempt to develop a uni- fied account. Rashal and colleagues (2022) demonstrated This study seems to indicate an interaction between top-down that top-down guidance of attention via an endogenous cue control and SL, where, when one mechanism is at a play, the diminished the benefit of target salience and the interference influence of the other is reduced or even abolished. In particu- from a salient distractor. Similarly, top-down control seems to lar, the collected behavioural results suggest that when strong prevail over the other AC signal, i.e., SL, in this experiment. top-down attention control is available, SL could not emerge, However, our EEG data showed that SL can block the benefit as if the information coming from the cue and guiding atten- of a valid cue in the high-frequency location, if in that loca- tion to the indicated spatial location bypasses the information tion the neural activity already reached the highest possible coming from the implicit learning process. Indeed, the fully peak due to the probability distribution of target frequency. reliable valid cue allowed participants to pre-allocate their As mentioned above, since the present work and the work attentional resources to the upcoming target location before of Rashal and colleagues (2022) used the same experimental the stimuli array onset, thus fully optimizing subsequent tar- task, it is possible to formulate hypotheses on the functional get selection. Nevertheless, our EEG results suggest that SL architecture of visual spatial attention. In particular, together was not totally overridden, at least at some stages of target these findings seem to suggest that there is a general domi- selection, and in turn being able to reduce the impact of top- nance of the mechanism underlying top-down allocation down guidance in some cases. of attention, since the effects of both bottom-up capture by Author note This study is part of a collaborative project (MAC-Brain: salience and SL on performance were diminished, or even Developing a Multi-scale account of Attentional Control as the con- absent, following a valid cue. In this scenario, one could straining interface between vision and action: A cross-species inves- argue that the final attentional choice is the result of the tigation of relevant neural circuits in the human and macaque Brain") funded under the European FLAG-ERA JTC 2017 program and associ- activity of only one mechanism, i.e., top-down control, that ated with the Human Brain Project. prevents the influence of all the others. Still, since in our experiment the cue was 100% valid and predicted the tar- Funding Open access funding provided by Università degli Studi di get location well in advance, participants could ignore the Verona within the CRUI-CARE Agreement. information coming from the target probability distribution Open practices statement Data will be made available on EBRAINS. when they already had certain and explicit information about Other materials for the experiments reported here are available upon where to find the target. As a consequence, the lack of effect request. None of the experiments was preregistered. of SL in the valid cue condition was likely not the result of a 1 3 Attention, Perception, & Psychophysics Carrasco, M. (2011). Visual attention: The past 25 years. Vision Declarations Research, 51(13), 1484–1525. https:// doi. org/ 10. 1016/j. visres. 2011. 04. 012 Conflict of interest The authors have no conflicts of interest to declare. Chelazzi, L., & Santandrea, E. (2018). The time constant of atten- tional control: Short, Medium and Long (Infinite?). Journal of Open Access This article is licensed under a Creative Commons Attri- Cognition, 1(1). Ubiquity Press. https://doi. or g/10. 5334/ joc. 24 bution 4.0 International License, which permits use, sharing, adapta- Chelazzi, L., Perlato, A., Santandrea, E., & Della Libera, C. (2013). tion, distribution and reproduction in any medium or format, as long Rewards teach visual selective attention. Vision Research, 85, as you give appropriate credit to the original author(s) and the source, 58–72. https:// doi. org/ 10. 1016/j. visres. 2012. 12. 005 provide a link to the Creative Commons licence, and indicate if changes Chelazzi, L., Eštočinová, J., Calletti, R., Gerfo, E. . Lo., Sani, I., were made. The images or other third party material in this article are Libera, C. Della., & Santandrea, E. (2014). Altering spatial included in the article's Creative Commons licence, unless indicated priority maps via reward-based learning. Journal of Neurosci- otherwise in a credit line to the material. If material is not included in ence, 34(25), 8594–8604. https://do i.o rg/1 0.15 23/J NEURO SCI. the article's Creative Commons licence and your intended use is not 0277- 14. 2014 permitted by statutory regulation or exceeds the permitted use, you will Desimone, R., & Duncan, J. (1995). Neural mechanisms of selective need to obtain permission directly from the copyright holder. To view a visual attention. Annual Review of Neuroscience, 18, 193–222. copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . Annual Reviews Inc. https:// doi. or g/ 10. 1146/ annur e v . ne. 18. 030195. 001205 Di Bello, F., Ben HadjHassen, S., Astrand, E., & Ben Hamed, S. References (2022). Prefrontal control of proactive and reactive mechanisms of visual suppression. Cerebral cortex (New York, N.Y. : 1991), 32(13), 2745–2761. https:// doi. org/ 10. 1093/ cercor/ bhab3 78 Aslin, R. N., & Newport, E. L. (2012). Statistical learning: From Druker, M., & Anderson, B. (2010). Spatial probability aids visual acquiring specific items to forming general rules. Current stimulus discrimination. Frontiers in Human Neuroscience, 4. Directions in Psychological Science, 21(3), 170–176. https:// https:// doi. org/ 10. 3389/ fnhum. 2010. 00063 doi. org/ 10. 1177/ 09637 21412 436806 Duncan, D., & Theeuwes, J. (2020). Statistical learning in the Awh, E., & Jonides, J. (2001). Overlapping mechanisms of attention absence of explicit top-down attention. Cortex, 131, 54–65. and spatial working memory. Trends in Cognitive Sciences, 5(3), https:// doi. org/ 10. 1016/j. cortex. 2020. 07. 006 119–126. Egeth, H. E., & Yantis, S. (1997). VISUAL ATTENTION: Control, Awh, E., Belopolsky, A. V., & Theeuwes, J. (2012). Top-down versus Representation, and Time Course. Annual Review of Psychol- bottom-up attentional control: A failed theoretical dichotomy. ogy, 48. www. annua lrevi ews. org Trends in Cognitive Sciences, 16(8), 437–443. https:// doi. org/ Eimer, M. (1994). An ERP study on visual spatial priming with 10. 1016/j. tics. 2012. 06. 010 peripheral onsets. Psychophysiology, 31(2), 154–163. https:// Bacigalupo, F., & Luck, S. J. (2019). Lateralized suppression of doi. org/ 10. 1111/j. 1469- 8986. 1994. tb010 35.x alpha-band EEG activity as a mechanism of target processing. Eimer, M. (1996). The N2pc component as an indicator of attentional Journal of Neuroscience, 39(5), 900–917. https:// doi. org/ 10. selectivity. Electroencephalography and clinical Neurophysiol- 1523/ JNEUR OSCI. 0183- 18. 2018 ogy, 99, 225–234. Baumgartner, H. M., Graulty, C. J., Hillyard, S. A., & Pitts, M. A. Failing, M., & Theeuwes, J. (2018). Selection history: How reward (2018). Does spatial attention modulate the earliest component modulates selectivity of visual attention. Psychonomic Bulletin of the visual evoked potential? Cognitive Neuroscience, 9(1–2), and Review, 25(2), 514–538. Springer New York LLC. https:// 4–19. https:// doi. org/ 10. 1080/ 17588 928. 2017. 13334 90 doi. org/ 10. 3758/ s13423- 017- 1380-y Beffara, B., Hadj-Bouziane, F., Hamed, S. B., Boehler, C. N., Ferrante, O., Patacca, A., Di Caro, V., Della Libera, C., Santandrea, Chelazzi, L., Santandrea, E. & Macaluso, E. (2022). Dynamic E., & Chelazzi, L. (2018). Altering spatial priority maps via sta- causal interactions between occipital and parietal cortex explain tistical learning of target selection and distractor filtering. Cor - how endogenous spatial attention and stimulus-driven salience tex, 102, 67–95. https:// doi. org/ 10. 1016/j. cortex. 2017. 09. 027 jointly shape the distribution of processing priorities in 2D vis- Fiser, J., & Aslin, R. N. (2001). Unsupervised statistical learning of ual space. Neuroimage, 255, 119206. https://doi. or g/10. 1016/j. higher-order spatial structures from visual scenes. Psychologi- neuro image. 119206 cal Science, 12(6), 499–504. Ben-Shachar, M. S., Lüdecke, D., & Makowski, D. (2020). effectsize: Folk, C. L., Remington, R. W., & Johnston, J. C. (1992). Involuntary Estimation of effect size indices and standardized parameters. covert orienting is contingent on attentional control settings. Journal of Open Source Software, 5(56), 2815. https:// doi. org/ Journal of Experimental Psychology. Human Perception and 10. 21105/ joss. 02815 Performance, 18(4), 1030–1044. https://doi. or g/10. 1037/ 0096- Bisley, J. W. (2011). The neural basis of visual attention. Journal 1523. 18.4. 1030 of Physiology, 589(1), 49–57. https://doi. or g/10. 1113/ jph ysiol. Gao, Y., & Theeuwes, J. (2020). Independent effects of statistical 2010. 192666 learning and top-down attention. Attention, Perception, and Bisley, J. W., & Goldberg, M. E. (2010). Attention, intention, and Psychophysics, 82(8), 3895–3906. https:// doi. or g/ 10. 3758/ priority in the parietal lobe. Annual Review of Neuroscience, 33, s13414- 020- 02115-x 1–21. https:// doi. org/ 10. 1146/ annur ev- neuro- 060909- 152823 Geng, J. J., & Behrmann, M. (2002). Probability cuing of target loca- Boudewyn, M. A., Luck, S. J., Farrens, J. L., & Kappenman, E. S. tion facilitates visual search implicitly in normal participants (2018). How many trials does it take to get a significant ERP and patients with hemispatial neglect. Psychological Science, effect? It depends. Psychophysiology, 55(6), e13049. https://d oi. 16, 520–525. org/ 10. 1111/ psyp. 13049 Geng, J. J., & Behrmann, M. (2005). Spatial probability as atten- Brunia, C. H. M., & van Boxtel, G. J. M. (2001). Wait and see. Inter- tional cue. Perception & Psychophysics, 67(7), 1252–1268. national Journal of Psychophysiology, 43(1), 59–75. https://d oi. https://link .spr ing er.com/ conte nt/ pdf/ 10. 3758/ BF031 93557. pdf org/ 10. 1016/ S0167- 8760(01) 00179-9 1 3 Attention, Perception, & Psychophysics Grent-‘t-Jong, T., & Woldorff, M. G. (2007). Timing and sequence of Newport, E. L., Hauser, M. D., Spaepen, G., & Aslin, R. N. (2004). brain activity in top-down control of visual-spatial attention. PLoS Learning at a distance: II. Statistical learning of non-adjacent Biology, 5(1), 12. https:// doi. org/ 10. 1371/ journ al. pbio. 00500 12 dependencies in a non-human primate. Cognitive Psychology, Heinze, H. J., Mangun, G. R., Burchert, W., Hinrichs, H., Scholz, M., 49(2), 85–117. Münte, T. F., Gös, A., Scherg, M., Johannes, S., Hundeshagen, Ngiam, W. X. Q., Adam, K. C. S., Quirk, C., Vogel, E. K., & Awh, E. H., Gazzaniga, M. S., & Hillyard, S. A. (1994). Combined spatial (2021). Estimating the statistical power to detect set size effects in and temporal imaging of brain activity during visual selective contralateral delay activity. Psychophysiology, 58, e13791. https:// attention in humans. Nature, 372(6506), 543–546. https://doi. or g/ doi. org/ 10. 1111/ psyp. 13791 10. 1038/ 37254 3a0 Noudoost, B., Chang, M. H., Steinmetz, N. A., & Moore, T. (2010). Ipata, A. E., Gee, A. L., Bisley, J. W., & Goldberg, M. E. (2009). Neu- Top-down control of visual attention. In Current Opinion in Neu- rons in the lateral intraparietal area create a priority map by the robiology, 20(2), 183–190. Elsevier Ltd. https://doi. or g/10. 1016/j. combination of disparate signals. Experimental Brain Research, conb. 2010. 02. 003 192(3), 479–488. https:// doi. org/ 10. 1007/ s00221- 008- 1557-8 Parisi, G., Mazzi, C., Colombari, E., Chiarelli, A. M., Metzger, B. A., Jiang, Y. V. (2018). Habitual versus goal-driven attention. Cortex, 102, Marzi, C. A., & Savazzi, S. (2020). Spatiotemporal dynamics of 107–120. Masson SpA. https://doi. or g/10. 1016/j. cor tex.2017. 06. attentional orienting and reorienting revealed by fast optical imag- 018 ing in occipital and parietal cortices. Neuroimage, 222, 117244. Kiss, M., Van Velzen, J., & Eimer, M. (2008). The N2pc component https:// doi. org/ 10. 1016/j. neuro image. 2020. 117244 and its links to attention shifts and spatially selective visual pro- Peirce, J. W. (2007). PsychoPy-Psychophysics software in python. cessing. Psychophysiology, 45(2), 240–249. https:// doi. org/ 10. Journal of Neuroscience Methods, 162(1–2), 8–13. https:// doi. 1111/j. 1469- 8986. 2007. 00611.xorg/ 10. 1016/j. jneum eth. 2006. 11. 017 Lawrence, M. (2015). ez: Easy analysis and visualization of factorial Posner, M. I. (1980). Orienting of attention. The Quarterly Journal of experiments [R package version 4.3.]. https:// github. com/ mike- Experimental Psychology, 32(1), 3–25. https:// doi. org/ 10. 1080/ lawre nce/ ez (Original work published 2011)00335 55800 82482 31 Leber, A. B., & Egeth, H. E. (2006). It’s under control: Top-down Ptak, R. (2012). The frontoparietal attention network of the human search strategies can override attentional capture. Psychonomic brain: Action, saliency, and a priority map of the environment. Bulletin & Review, 13(1), 132–138. https://d oi.o rg/1 0.3 758/b f031 Neuroscientist, 18(5), 502–515. h tt ps :/ / d oi . o rg/ 1 0 . 1 17 7/ 1 0 73 8 9382458411 409051 Liebrand, M., Pein, I., Tzvi, E., & Krämer, U. M. (2017). Tempo- R Core Team. (2016). R: A language and environment for statistical ral dynamics of proactive and reactive motor inhibition. Fron- computing. https:// www. gbif. org/ tool/ 81287/r- a- langu age- and- tiers in Human Neuroscience, 11, 204. https://doi. or g/10. 3389/ envir onment- for- stati stical- compu ting fnhum. 2017. 00204 Rashal, E., Senoussi, M., Santandrea, E., Ben-Hamed, S., Macaluso, Liesefeld, H. R., Liesefeld, A. M., Töllner, T., & Müller, H. J. (2017). E., Chelazzi, L., & Boehler, C. N. (2022). An EEG study of the Attentional capture in visual search: Capture and post-capture combined effects of top-down and bottom-up attentional selection dynamics revealed by EEG. NeuroImage, 156, 166–173. https:// under varying task difficulty. Psychophysiology. https:// doi. org/ doi. org/ 10. 1016/j. neuro image. 2017. 05. 01610. 1111/ psyp. 14002 Livingstone, A. C., Christie, G. J., Wright, R. D., & McDonald, J. Reynolds, J. H., & Chelazzi, L. (2004). Attentional modulation of J. (2017). Signal enhancement, not active suppression, follows visual processing. Annual Review of Neuroscience, 27, 611–647. the contingent capture of visual attention. Journal of Experi-https:// doi. org/ 10. 1146/ annur ev. neuro. 26. 041002. 131039 mental Psychology: Human Perception and Performance, 43(2), Reynolds, J. H., & Heeger, D. J. (2009). The normalization model 219–224. of attention. Neuron, 61(2), 168–185. https:// doi. org/ 10. 1016/j. Luck, S. J., & Hillyard, S. A. (1994). Spatial filtering during visual neuron. 2009. 01. 002 search: Evidence from human electrophysiology. Journal of Rosa-Salva, O., Fiser, J., Versace, E., Dolci, C., Chehaimi, S., Santo- Experimental Psychology: Human Perception and Performance, lin, C., & Vallortigara, G. (2018). Spontaneous learning of visual 20(5), 1000–1014. structures in domestic chicks. Animals, 8(8), 135. https://doi. or g/ Luck, S. J., Girelli, M., Mcdermott, M. T., & Ford, M. A. (1997). 10. 3390/ ani80 80135 Bridging the gap between monkey neurophysiology and human Saffran, J. R., Aslin, R. N., & Newport, E. L. (1996). Statistical learn- perception: An ambiguity resolution theory of visual selective ing by 8-month-old infants. Cognition, 274, 1–10. attention. Cognitive Psychology, 33, 64–87. Santolin, C., Rosa-Salva, O., Vallortigara, G., & Regolin, L. (2016). Luck, S. J., Woodman, G. F., & Vogel, E. K. (2000). Event-related Unsupervised statistical learning in newly hatched chicks. Cur- potential studies of attention. Trends in Cognitive Sciences, rent Biology, 26(23), R1218–R1220. https:// doi. org/ 10. 1016/j. 4(11), 432–440. h t t p s : / / d o i . o r g / 1 0 . 1 0 1 6 / S 1 3 6 4 - 6 6 1 3 ( 0 0 ) cub. 2016. 10. 011 01545-X Schevernels, H., Krebs, R. M., Santens, P., Woldorff, M. G., & Boe- Macaluso, E., & Doricchi, F. (2013). Attention and predictions: hler, C. N. (2014). Task preparation processes related to reward Control of spatial attention beyond the endogenous-exogenous prediction precede those related to task-difficulty expectation. dichotomy. Frontiers in Human Neuroscience (OCT). Frontiers NeuroImage, 84, 639–647. https://doi. or g/10. 1016/j. neur oimag e. Media S. A. https:// doi. org/ 10. 3389/ fnhum. 2013. 006852013. 09. 039 Mangun, G. (1995). Neural mechanisms of visual selective attention. Serences, J. T., & Yantis, S. (2007). Spatially selective representations Psychophysiology, 32, 4e18. of voluntary and stimulus-driven attentional priority in human Mangun, G. R., & Hillyard, S. A. (1991). Modulations of sensorye- occipital, parietal, and frontal cortex. Cerebral Cortex, 17(2), voked brain potentials indicate changes in perceptual processing 284–293. https:// doi. org/ 10. 1093/ cercor/ bhj146 during visual-spatial priming. Journal Experimental Psychology Tecce, J. J. (1972). Contingent negative variation (CNV) and psycho- Human, 17, 1057e1074. logical processes in man. Psychological Bulletin, 77(2), 73–108. Munneke, J., Heslenfeld, D. J., & Theeuwes, J. (2010). Spatial work-https:// doi. org/ 10. 1037/ h0032 177 ing memory effects in early visual cortex. Brain and Cognition, Theeuwes, J. (1991). Exogenous and endogenous control of attention: 72(3), 368–377. The effect of visual onsets and offsets. Perception & Psychophys- ics, 49(1), 83–90. 1 3 Attention, Perception, & Psychophysics Theeuwes, J. (1994). Stimulus-driven capture and attentional set: Walter, W. G., Cooper, R., Aldridge, V. J., McCallum, W. C., & Win- Selective search for color and visual abrupt onsets. Journal of ter, A. L. (1964). Contingent negative variation: An electric sign Experimental Psychology: Human Perception and Performance, of sensorimotor association and expectancy in the human brain. 20(4), 799–806. https:// doi. org/ 10. 1037/ 0096- 1523. 20.4. 799 Nature., 203(4943), 380–384. https:// doi. org/ 10. 1038/ 20338 0a0 Theeuwes, J. (2010). Top-down and bottom-up control of visual selec- Wang, B., van Driel, J., Ort, E. & Theeuwes, J. (2019). Anticipatory tion. Acta Psychologica, 135(2), 77–99. https://doi. or g/10. 1016/j. distractor suppression elicited by statistical regularities in visual actpsy. 2010. 02. 006 search. Journal of Cognitive Neuroscience, 1–14. https:// doi. org/ Theeuwes, J., & Godijn, R. (2004). Inhibition-of-return and oculomotor 10. 1162/ jocn_a_ 01433 interference. Vision Research, 44(12), 1485–1492. https://doi. or g/ Woodman, G. F., & Luck, S. J. (2003). Serial deployment of attention 10. 1016/j. visres. 2003. 09. 035 during visual search. Journal of Experimental Psychology. Human Van Den Berg, B., Krebs, R. M., Lorist, M. M., & Woldorff, M. G. Perception and Performance, 29(1), 121–138. https:// doi. org/ 10. (2014). Utilization of reward-prospect enhances preparatory 1037// 0096- 1523. 29.1. 121 attention and reduces stimulus conflict. Cognitive, Affective and Wu, C.-T., Libertus, M. E., Meyerhoff, K. L., & Woldorff, M. G. Behavioral Neuroscience, 14(2), 561–577. https:// doi. or g/ 10. (2011). The temporal dynamics of object processing in visual 3758/ s13415- 014- 0281-z cortex during the transition from distributed to focused spatial Van Moorselaar, D., & Slagter, H. A. (2019). Learning what is irrel- attention. J. Cognitive Neuroscience, 23, 4094–4105. evant or relevant: Expectations facilitate distractor inhibition and Yantis, S., & Egeth, H. E. (1999). On the distinction between vis- target facilitation through distinct neural mechanisms. Journal of ual salience and stimulus-driven attentional capture. Journal of Neuroscience, 39(35), 6953–6967. https:// doi. or g/ 10. 1523/ JNEUR Experimental Psychology: Human Perception and Performance, OSCI. 0593- 19. 2019 25(3), 661–676. https:// doi. org/ 10. 1037/ 0096- 1523. 25.3. 661 Van Voorhis, S., & Hillyard, S. A. (1977). Visual evoked potentials and Zivony, A., Allon, A. S., Luria, R., & Lamy, D. (2018). Dissociating selective attention to points in space. Perception & Psychophysics, between the N2pc and attentional shifting: An attentional blink 22(1), 54–62. https:// doi. org/ 10. 3758/ bf032 06080 study. Neuropsychologia, 121, 153–163. https://doi. or g/10. 1016/j. Verleger, R., Wascher, E., Wauschkuhn, B., Jaśkowski, P., Allouni, B., neuro psych ologia. 2018. 11. 003 Trillenberg, P., & Wessel, K. (1999). Consequences of altered cer- ebellar input for the cortical regulation of motor coordination, as Publisher’s note Springer Nature remains neutral with regard to reflected in EEG potentials. Experimental brain research, 127(4), jurisdictional claims in published maps and institutional affiliations. 409–422. https:// doi. org/ 10. 1007/ s0022 10050 809 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Attention Perception & Psychophysics Springer Journals

Integrated effects of top-down attention and statistical learning during visual search: An EEG study

Loading next page...
 
/lp/springer-journals/integrated-effects-of-top-down-attention-and-statistical-learning-daZX4qjga0

References (84)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2023
ISSN
1943-3921
eISSN
1943-393X
DOI
10.3758/s13414-023-02728-y
Publisher site
See Article on Publisher Site

Abstract

The present study aims to investigate how the competition between visual elements is solved by top-down and/or statistical learning (SL) attentional control (AC) mechanisms when active together. We hypothesized that the “winner” element that will undergo further processing is selected either by one AC mechanism that prevails over the other, or by the joint activity of both mechanisms. To test these hypotheses, we conducted a visual search experiment that combined an endogenous cueing protocol (valid vs. neutral cue) and an imbalance of target frequency distribution across locations (high- vs. low-frequency location). The unique and combined effects of top-down control and SL mechanisms were measured on behaviour and amplitudes of three evoked-response potential (ERP) components (i.e., N2pc, P1, CNV) related to attentional processing. Our behavioural results showed better performance for validly cued targets and for targets in the high-frequency location. The two factors were found to interact, so that SL effects emerged only in the absence of top-down guidance. Whereas the CNV and P1 only displayed a main effect of cueing, for the N2pc we observed an interaction between cueing and SL, revealing a cueing effect for targets in the low-frequency condition, but not in the high-frequency condition. Thus, our data support the view that top-down control and SL work in a conjoint, integrated manner during target selection. In particular, SL mechanisms are reduced or even absent when a fully reliable top-down guidance of attention is at play. Keywords N2pc · P1 · Statistical learning · Endogenous cueing · Attention control · Priority map Introduction select the relevant information and tune out what is irrel- evant (Desimone & Duncan, 1995; Reynolds & Chelazzi, In everyday life, the large number of visual inputs com- 2004). This process involves one or multiple attentional con- ing from the environment greatly exceeds our sensory and trol (AC) mechanisms that assign attentional priority to a cognitive processing capacities. Looking for a book in a certain stimulus or location in the visual field. A prominent crowded library can be a difficult task since, at all times, theory of attentional guidance is the priority map theory, all the available visual stimuli compete with each other in which suggests a neural representation of visual space that is order to gain access to further processing. Visual attention topographically organized (Bisley & Goldberg, 2010; Ptak, is the cognitive function that acts as a filter, allowing us to 2012). Depending on the context and time, each location in the visual space is suggested to have a specific level of neuronal activity that is determined by the amount of atten- * Carola Dolci tional priority assigned to that location (Awh et al., 2012; carola.dolci@univr.it Chelazzi et al., 2013; Di Bello et al., 2022; Ipata et al., 2009; Department of Neuroscience, Biomedicine, and Movement Serences & Yantis, 2007). The highest activation peak trig- Science, University of Verona, Strada le Grazie, 8, gers a winner-takes-all process, leading to the target at that 37134 Verona, Italy location being selected (Bisley, 2011; Chelazzi et al., 2014; Department of Experimental Psychology, Ghent University, Macaluso & Doricchi, 2013; Noudoost et al., 2010). Thus, Ghent, Belgium the distribution of attentional resources within the prior- Institut des Sciences Cognitives Marc-Jeannerod, Lyon, ity map would be influenced by the action of different AC France mechanisms. Lyon Neuroscience Research Center, Lyon, France Vol.:(0123456789) 1 3 Attention, Perception, & Psychophysics Individual priority signals may originate from various versus mechanism prevalence in solving the competition sources. Traditionally, they have been separated into two between stimuli. Previous studies are more in line with the main categories: top-down and bottom-up. Top-down (or first hypothesis, arguing that these two mechanisms operate goal-directed) AC is an endogenous process, driven by independently from each other, with the influence of the active volitional selection of items that are relevant to a per- two adding up in a linear manner when engaged at the same son’s goals or instructions (Carrasco, 2011; Egeth & Yantis, time (Duncan & Theeuwes, 2020; Gao & Theeuwes, 2020; 1997; Leber & Egeth, 2006; Parisi et al., 2020; Reynolds Geng & Behrmann, 2005). For instance, Gao and Theeu- & Heeger, 2009). For instance, the competition between wes (2020) showed how SL biased the competition in favour stimuli can be solved by the presence of a central visual cue, of a target that appeared frequently in a ceratian location, which indicates the forthcoming target location and allows compared with a target that appeared in a rare location in the pre-allocation of attentional resources to that position, the array, and that this effect was not affected by top-down facilitating target detection (Posner, 1980). In contrast, bot- attention being directed to one or the other location. At the tom-up attention is an exogenous AC mechanism by which same time, better performance was found when participants attentional resources are automatically allocated toward a could benefit from valid (vs. invalid) information that was salient stimulus with highly noticeable feature properties, given to the top-down AC mechanism, when the target was such as luminance, color, or shape (Theeuwes, 1991, 1994; presented in both the high- and the low-frequency locations Theeuwes & Godijn, 2004; Yantis & Egeth, 1999). (Gao & Theeuwes, 2020). This suggests that both mecha- In recent years, it has been shown that people can implic- nisms can guide independently the attentional selection of itly develop another type of bias specifically linked to the specific spatial locations on the priority map. individual’s previous experience with a given context and/ However, in that study, top-down control was always pre- or stimuli, which can also guide target selection (Awh et al., sent, as participants were instructed to attend to a location in 2012; Ferrante et al., 2018; Jiang, 2018). Therefore, a third the array that could correspond to the location of the upcom- AC category has been introduced: experience-dependent ing target, but not with complete certainty; only on 50% of AC (Awh et al., 2012; Chelazzi & Santandrea, 2018; Fail- the trials the cue indicated the exact target location, whereas ing & Theeuwes, 2018). One of the experience-dependent on the other half of the trials it indicated a location nearby. mechanisms is statistical learning (SL), which allows Thus, it is possible that the cue validity led to the absence of humans, but also other animals, to implicitly extract regu- an observable interaction between the two mechanisms, as larities from the environment even without having explicit the level of uncertainty introduced in the paradigm may have instructions (Aslin & Newport, 2012; Druker & Anderson, prevented a possible interaction between the mechanisms. 2010; Ferrante et al., 2018; Duncan & Theeuwes, 2020; For this reason, in the current study we provided participants for evidence in non-human primates: Newport et al., 2004; with a fully predictive top-down control guidance, by using and chicken: Rosa-Salva et al., 2018; Santolin et al., 2016). a 100% valid cue that pointed to the upcoming target loca- In particular, the probability with which a target element tion and compared it to a condition where top-down control occurs in a specific location was found to induce an atten- was not at a play, where participants were provided with an tional bias toward the location where it is more likely to uninformative neutral cue. appear, without the participant being consciously aware of this (Geng & Behrmann, 2002, 2005). EEG markers of visual selective attention As described above, many studies investigated how visual attention is guided by individual AC mechanisms, but they To further examine the prioritization process in visual do not specify how attentional selection is reconfigured search, in the present study, we investigated the neural when different attentional biases are at a play. Indeed, in mechanism underlying it. Specifically, we focused on three many aspects of everyday life multiple AC mechanisms can well-established evoked-response potential (ERP) compo- act simultaneously, and it is still unclear how they inter- nents related to attentional selection: the cue-related contin- act with one another in the prioritization process and how gent negative variation (CNV), and the target-related P1 and the final attentional choice is established. One possibility is N2pc. When investigating selective attention using visual- that, when acting simultaneously, the activity of all the AC search paradigms, the main EEG marker of interest is the mechanisms is added-up and they jointly contribute to solv- N2pc, which is the negative deflection at posterior electrodes ing the competition in favour of one stimulus. Alternatively, contralateral to the target, typically emerging around 200 one mechanism may prevail over the other, thus exclusively ms from the onset of a lateralized target. Traditionally, the governing target selection. N2pc has been assumed to index the shift of covert attention The aim of the present study was to study the combination toward a task-relevant, or salient, stimulus (Eimer, 1996; of AC mechanisms, specifically, top-down AC and statistical Luck & Hillyard, 1994), but other findings suggest that the learning, and to test the hypotheses of a joint contribution N2pc reflects various aspect of target processing (Kiss et al., 1 3 Attention, Perception, & Psychophysics 2008; Theeuwes, 2010; Zivony et al., 2018). Relevant for visual stimuli appeared on the cued compared with the non- the current study, in a recent work using a very similar task cued side of the array, suggesting that it is an early manifes- to the one used here, Rashal and colleagues (2022; Experi- tation of top-down attentional control (Eimer, 1994; Mangun ment 2) observed an N2pc for targets preceded by a valid & Hillyard, 1991; Van Voorhis & Hillyard, 1977). endogenous cue to the target location, suggesting that the N2pc reflected attentional processes also following topdown Aim and hypotheses of the study deployment of attention to that location. As SL induces a change of attentional priority in favour We devised a visual search task to investigate both isolated of the high-frequency target location, it might be expected and integrated effects of different sources of AC during the that a facilitation of target selection (Ferrante et al., 2018; target selection process. In particular, we focused on top- Geng & Behrmann, 2002)  would be accompanied by a down attention control, which we manipulated via endog- larger N2pc elicited by that target. Still, a recent study enous cueing, and statistical learning, which was manipu- by van Moorselaar and Slagter (2019) found instead a lated by an imbalance of target frequency across locations. reduction of N2pc amplitudes when the target appeared By comparing performance in trials where targets appeared frequently in a certain location. In their study, however, in the high- (HFTL) versus low- frequency target location the target competed with only one other stimulus (i.e., the (LFTL) and were preceded by an informative (valid) or distractor), making the task easier as attentional selection non-informative (neutral) cue, we tested whether top-down was quickly accomplished (see also Rashal et al., 2022, for control and SL, when active together, both contribute to evidence that the N2pc is modulated by difficulty-related assigning attentional priority to a specific spatial location factors). Furthermore, statistical learning in that study was (hypothesis 1) or if one mechanism is blocked by the pres- constrained to target repetitions, with the target appearing ence of the other mechanism (hypothesis 2). Specifically, if at the same location for a number of consecutive trials (4 the two mechanisms operate independently, better perfor- trials) within a sequence, but that location varied across mance and a larger N2pc should be observed for targets in the duration of the experiment. Critically, the modulation the HFTL compared with the LFTL condition irrespective of the N2pc reported by van Moorselaar and Slagter (2019) of the cueing condition. At the same time, cueing effects revealed that the N2pc amplitude, and thus the deployment should emerge regardless of the target location frequency of attentional resources needed for target selection, was condition, and better performance and a larger N2pc should diminished for repeating target location in consecutive tri- emerge following a valid cue when the target appears in both als. That is, the N2pc was larger in the first trial than in the the HFTL and the LFTL (for behavioural evidence, see Gao last trial of the repetition sequence. However, this result & Theeuwes, 2020). Alternatively, if the two mechanisms may be attributed to inter-trial priming and may not apply to interact with each other, we should find that one mechanism a situation where SL is established across the entire experi- affects the other in some way. For example, it is possible ment. As a matter of fact, in classic SL paradigms, target that top-down control blocks the effect of SL, such that its location frequency is associated with just one (or a few) effect can be reduced or even gated by pre-cueing the target spatial location(s) or region(s) across the entire experiment, location, resulting in a smaller difference in performance allowing SL to be reinforced continuously and inducing an and N2pc mean amplitude between targets in the HFTL attentional enhancement in favour of that location. and LFTL following a valid cue compared with the same Two other components related to attentional control are difference in performance and N2pc amplitudes for targets the post-cue CNV and the post-target P1 (Mangun, 1995; following a neutral cue. Similarly, it can be that SL blocks Schevernels et al., 2014; Van Den Berg et al., 2014), the first top-down control. In this case, we should find that the benefit specifically related to top-down control, while the latter is of validly cueing the target location is reduced by the pres- potentially modulated by top-down and bottom-up mecha- ence of target-location frequency imbalance, resulting in a nisms. The CNV is characterized by a slow, negative-going smaller cueing effect on behaviour and N2pc amplitude in waveform normally detected in central areas after the pres- the HFTL compared with the LFTL conditions. entation of a warning stimulus such as a cue (Walter et al., Additionally, we examined two other EEG components 1964), likely reflecting a general preparatory attention dur - mostly related to the top-down control: the P1 during vis- ing the cue-target interval of attentional tasks (e.g., Grent- ual search, and the CNV during the cue-target interval. By ‘t-Jong & Woldorff, 2007). The P1 is the first positive-going looking at the CNV and P1 components, we can investigate ERP component, starting around 90 ms after target-array modulations to top-down attentional orienting pre- and post- onset, and displays increased amplitudes over the occipital stimulus array onset. A larger CNV should emerge follow- scalp contralateral to the precued location (Baumgartner ing a valid compared with a neutral cue, reflecting advance et al., 2018; Mangun & Hillyard, 1991). P1 amplitudes have preparation for selecting the target stimulus (Rashal et al., been demonstrated to be enhanced when the corresponding 2022; Schevernels et al., 2014; Van Den Berg et al., 2014). 1 3 Attention, Perception, & Psychophysics Furthermore, the P1 could also be modulated by the pres- on a given trial, with all stimuli being drawn in the same ence or absence of a valid cue. Specifically, we expected a colour. The choice for two colours, which was not essen- larger P1 following a valid compared with a neutral cue, tial to the present task, and which was fully counterbal- indicating an early stimulus categorization when a stimulus anced across conditions, largely relates to earlier work of is presented in the expected spatial location (Heinze et al., ours using the same global approach (Rashal et al., 2022). 1994; Mangun, 1995). Indeed, Livingstone and colleagues Within each stimulus, there was a small gap (diameter of (2017) demonstrated that P1 indexes an enhanced processing 0.25°) of the same grey colour as the background and posi- for the search item pointed by a valid cue at a stage of vision tioned at the upper or lower part. The target was a bar tilted that precedes attentional selection. ±25° across the vertical axis, whereas the other stimuli that Lastly, even if a modulation of the CNV and P1 have been had to be ignored (distractors) were bars tilted ±25° across most clearly associated with cueing, here we tested if SL was the horizontal axis. Two stimuli were presented in the upper able to affect the general preparation and attentional orient- visual field, two on the horizontal midline and two in the ing pre- and post-stimulus array onset. If so, as for N2pc, we lower visual field (Fig.  1, panel a). Since evidence indicates would expect a larger CNV and P1 for targets in the HFTL, that the N2pc is usually larger in the lower visual field and compared with the LFTL condition, and this effect could on the horizontal meridian than in the upper visual field interact with the cueing manipulation. (Bacigalupo & Luck, 2019; Luck et al., 1997), the target never appeared in the two upper locations, which hence just contained filler items. In each visual search display, Methods six stimuli were presented, centred equidistantly 7° away from a white fixation cross (0.5°×0.5°; RGB: 255, 255, 255; Participants luminance 190.2 cd/m ). Before the onset of the stimulus array, a cue stimulus Twenty-four healthy volunteers (four males; mean age 23.62 was presented around the fixation cross (Fig.  1, panel a). years, SD ±3.4 years) with normal or corrected-to-normal The cue consisted of a geometric shape (dimension: 1.5° visual acuity participated in this experiment. None of them × 1.5°) made up of six separate corners, each pointing at had previously taken part in similar or related studies, and one of the stimulus locations. In the case of the neutral cue, they were naive to the purpose of the present study. At the all the corners were coloured with the same pink (RGB: end of the experiment, they received a fixed monetary com- 120, 0, 90; luminance: 89.5 cd/m ), whereas in the case pensation for their participation (€32.5). All subjects gave of the valid cue, five corners were pink and one was cyan their written informed consent before participation. The blue (RGB: 0, 56, 158; luminance: 81.1 cd/m ), indicat- present study was approved by the ethics committee of the ing in which spatial position the target element would be Faculty of Psychology and Educational Sciences of Ghent presented (Fig. 1, panel a). University (code 2021/09). Experimental design Apparatus and stimuli A central cue presented prior to the target array onset was The experiment was conducted in a dimly lit and quiet room, either valid or neutral. In the valid cue condition, the loca- where participants sat in front of a 24-in. Benq XL2411Z tion of the upcoming target was predicted with a validity of LED monitor controlled by a Dell Optiplex 9020 tower 100%. In the neutral cue condition, the cue did not include with Intel Core i5-4590 processor, at 60-Hz refresh-rate. information about the target location. Importantly, in order The viewing distance was held constant at 60 cm by using to not mix the two AC manipulations, SL was manipulated an adjustable chin rest. The experiment was run with the exclusively following neutral cues by introducing, unbe- PsychoPy (v1.84.2) software (Peirce, 2007). Good central known to the participants, an imbalance of target frequency fixation by the participants was monitored using the cam - appearance across the four possible target locations: high, era of an Eyelink 1000 plus (SR Research, Canada). The low and two intermediate location frequencies (Ferrante experimenter was sitting in a different room and warned et al., 2018). The valid cue, when present, indicated with the participants during breaks in case eye-movements were equal frequency each of the four possible target locations (96 observed in the preceding block, to allow correction. trials each location). The neutral cue trials (1,216 trials; 76% The stimuli were rectangular bars of 2.0° × 0.5° in size, of all trials) provided a baseline where the SL effect could be either green (RGB coordinates: 0, 86, 0; luminance: 138.5 assessed in the absence of top-down guidance. Here the tar- cd/m ) or red (RGB values: 170, 0, 0; luminance: 64.8 cd/ get appeared in the high-frequency location 50% of the trials m ), presented on a grey background (RGB: 128, 128, 128; (608 trials), in the low-frequency location for 7.9% of the luminance: 85.5 cd/m ). Colours were randomly chosen trials (96 trials), and in each of the intermediate-frequency 1 3 Attention, Perception, & Psychophysics Fig. 1 a Examples of the trial sequence. Top row: a neutral cue pre- and was the bar tilted ±25° from its vertical axis, while the non-tar- ceded target array onset. Bottom row: target location was predicted by gets were tilted ±25° from their horizontal axis. b Target frequency a valid cue. The target is indicated in the figure by the dashed circle distribution across groups (during neutral cue trials only). Note that (for illustration purposes; no such circle was present during the task) the target never appeared in the two upper locations locations for 21% of these trials (256 trials each). We did duration of the experiment. After a random interval jittered not introduce an imbalance of target frequency appearance between 100 and 300 ms, the cue appeared for 480 ms. After in the valid cue condition because doing so would mean a cue-target interval (CTI), jittered between 700 and 900 ms, introducing an imbalance of valid cues. This, in turn, would the search display appeared and remained visible for 300 complicate the interpretation of the results, as it would be ms. Responses were recorded from the onset of the search impossible to disentangle the benefit in target detection due display until 1,200 ms after display offset, for a total of 1,500 to SL of the target location, or SL of the valid cue, or both. ms. Afterwards, a new trial sequence started automatically. Participants were randomly assigned to one of four groups The task was to discriminate the position of the gap within (Fig. 1, panel b), each with a different spatial configuration the target item (top vs. bottom) by pressing the letter ‘M’ on of target-location probabilities, but with the constraint that the keyboard with their right index finger if the gap was in the high-probability and low-probability conditions were the lower part, or the letter ‘Z’ with their left index finger always in opposite locations in the left versus right visual if it was in the upper part. The experiment included a total field. of 1,600 trials, divided into eight blocks. Before starting the actual experiment, a practice phase of 64 trials was used to Procedure allow participants to familiarise themselves with the task. All the conditions previously described were presented in a Each experimental trial (Fig. 1, panel a) started with a fix- fully randomized order. Participants were instructed to main- ation cross, which remained on the screen for the whole tain their eyes on the fixation cross, and fixation quality was 1 3 Attention, Perception, & Psychophysics monitored by the experimenter by means of the online eye- Liesefeld et al., 2017; Rashal et al., 2022). To determine position display of the eye-tracker. the analysis time-windows for each of these EEG markers, In order to evaluate if participants were aware of the fre- we took the canonical values used in the literature: for the quency manipulation, a survey was conducted at the end of CNV, we selected a time-range from 700 ms after cue onset the experiment (see Ferrante et al., 2018). Participants were until approximately the earliest point in the CTI in which first asked to report whether they noticed something about the search display could appear (plus 70 ms, accounting the spatial distribution of target stimuli, and in case they for transduction delay into visual cortex), i.e., 700–1,250 responded “yes”, they had to report (or guess) the locations ms (e.g., Liebrand et al., 2017; Rashal et al., 2022). For where the target was presented most frequently. the N2pc and P1, the respective time-ranges were set to 200–300 ms (N2pc) and 90–140 ms (P1) after the search- Electrophysiological recording and analysis display onset, in line with the existing literature (e.g., Eimer, 1996; Luck et al., 2000; Mangun & Hillyard, 1991). Note EEG data were recorded using a Brain Products actiCHamp that counter to most of the earlier N2pc and P1 literature, the 64-channel system (Brain Products, Gilching, Germany) target location frequency imbalance led to the fact that for a with 64 active scalp electrodes positioned according to the given participant, the contralateral and ipsilateral locations standard international 10–10 system. Signals were recorded were either PO7 or PO8, and could not be collapsed across at a 500-Hz sampling rate, using Fz as the online reference those locations for different conditions, with corresponding and then re-referenced offline to the average of TP9 and targets on the left and right (e.g., Wu et al., 2011). Therefore, TP10, corresponding to the left and right mastoids. Fz was in this study, the average across locations was possible only then restored to the dataset. A high-pass filter of 0.1 Hz across groups (Wang et al., 2019). was applied to the raw data and segments of the continuous Analyses were performed using R 3.6.2 (R Core Team, data, with clearly identifiable, large artefacts (not including 2016) with ez (Lawrence, 2011/2015) and effectsize blinks and eye movements) were excluded by manual inspec- (Ben-Shachar et al., 2020) packages. For CNV we used tion. Successively, independent component analysis (ICA) rm-ANOVAs to compare the mean amplitude in the dif- was used to remove components related to eye blinks and ferent conditions, whereas for N2pc and P1 we first cal- (residual) eye movements. We then segmented the data into culated the mean amplitude of the ipsi and contra location epochs from −200 ms to 2,900 ms relative to cue onset and of interest, and then performed rm-ANOVAs to compare from −200 ms to 800 ms relative to the stimuli array onset. the difference waves (DWs) resulting from the subtrac- We then baseline-corrected with regard to the 200-ms pre- tion contra-minus-ipsi between different conditions. All cue or pre-stimuli period, respectively. Then, a second arti- these analyses were performed only using trials with cor- fact rejection (AR) was performed to flag and remove epochs rect responses. P values were corrected with Greenhouse- that contained artefacts in the analysed channels (PO7/8; Geisser epsilon in cases of significant sphericity violation. absolute amplitude exceeding ±100 μV). On average this led to exclusion less than 3% of the total trials. In order to study the temporal dynamics of attentional Results orienting and subsequent visual search, we focused on three components, namely the cue-evoked CNV and the P1 and Behaviour N2pc elicited by the search array. The CNV was examined at Cz using the cue-locked epochs (Verleger et al., 1999; In order to assess the effects of and interaction between sta- Rashal et al., 2022), whereas for the N2pc and P1 we used tistical learning and top-down mechanisms, 2 × 2 rm-ANO- the average of two electrodes capturing activity at PO7/ VAs were conducted with Target Location Frequency (high, PO8, where the N2pc and P1 are usually the largest (e.g., low) and Cue (valid, neutral) for accuracy and reaction time (RT). These analyses showed significant main effects of Cue for accuracy and RT [ACC: F(1, 23) = 15.32, p = 0.0006, 2 2 η = 0.39; RT: F(1, 23) = 134.41, p < 0.0001, η = 0.85], p p In order to maintain the same experimental protocol as in our previ- and Target Location Frequency for RT [ACC: F(1, 23) = ous related work, the cap was placed slightly further to the back than 2 2 0.46, p = 0.50, η = 0.01; RT: F(1, 23) = 6.10, p = 0.02, η p p typical, positioning FCz at the Cz site (Rashal et al., 2022). Thus, to = 0.20]. Importantly, a significant interaction between the test the N2pc and P1 components we report the results from the aver- two factors was observed for RT [ACC: F(1, 23) = 0.71, p = age between P3/P5 and P4/P6, which are the channel pairs that were 2 2 closest (~1 cm off) to the PO7/PO8 sites. Similarly, for the CNV the 0.40, η = 0.03; RT: F(1, 23) = 9.32, p = 0.005, η = 0.29]. p p channel used was FCz, which directly corresponds to Cz on the scalp Post hoc paired t-tests (two-tailed) revealed that participants (Verleger et al., 1999; Rashal et al., 2022). For simplicity, we refer to were significantly faster in detecting the target in the HFTL the actual scalp locations, rather than the labels on the cap. compared with the LFTL, but only when the cue was neutral 1 3 Attention, Perception, & Psychophysics Fig. 2 Mean accuracies (left) and reaction times (RTs; right) as a function of cue and target frequency conditions. Error bars represent the standard error of the mean [t(23) = -2.84; p = 0.009, Cohen’s d = -0.34, -30 ms], and the analysis did not change the main results, corroborating not when the cue was valid [t(23) = -0.73; p = 0.47, Cohen’s the implicit nature of the learning process. d = -0.04, -3 ms], suggesting that top-down control is able to exert a gating effect over SL mechanisms. Furthermore, N2pc the benefit of the valid cue was observed in both the HFTL [t(23) = 12.25; p < 0.0001, Cohen’s d = 1.02, 100 ms] and To investigate the effect of top-down AC and SL on the the LFTL [t(23) = 9.75; p < 0.0001, Cohen’s d = 1.49, 130 N2pc, we first investigated if the component was present ms] conditions (Fig. 2), but with a larger benefit for the cue in each condition. One-sample t-tests (one-tailed) showed in the latter (130 vs. 100 ms). a marginally significant N2pc (mean amplitude lesser than At the end of the experimental session, four participants zero) for targets at the HFTL (following neutral cue: t(23) reported having noticed something peculiar regarding the = −1.59, p = 0.06, Cohen's d = −0.32; following valid cue: target frequency, and identified the correct high-frequency t(23) = −1.57, p = 0.06, Cohen's d = −0.32), but not for spatial location as the location where the target was more targets at the LFTL (following neutral cue: t(23) = 0.25, p = likely to appear. However, excluding these participants from 0.59, Cohen's d = 0.05; following valid cue: t(23) = −0.90, p = 0.18, Cohen's d = −0.18). An rm-ANOVA was then conducted with Cue (valid, neutral) and Target Location Frequency (high, low). This analysis considered the contra-minus-ipsi difference waves, directly focusing on attentional lateralization effects. This To examine whether the SL effect was location-specific or, instead, analysis showed a significant main effect of Cue [F (1,23) if it was linked to a hemifield-based representation of space, we 2 = 5.84, p = 0.023, η = 0.20], but not of Target Location repeated the same analysis on the two intermediate locations that were associated with the hemisphere that contained the HTLF and LTLF of each participant. We performed a 2 × 2 rm-ANOVA with Cue (valid, neutral) and Target Location Frequency (intermediate- A 2 × 2 rm-ANOVA with Cue (valid, neutral) and Target Location high, intermediate-low). For accuracy, this analysis showed a signifi- Frequency (high, low) was performed excluding the four subjects who cant main effect of Cue [F(1, 23) = 10.43, p = 0.003, η = 0.31], reported having explicitly noticed the target frequency imbalance. but not of Target Location Frequency [F(1, 23) = 0.16, p = 0.69, For RTs, the analysis confirmed the two main effects [Cue: F(1,19) η = 0.006], and no significant interaction between the two factors = 115.04, p < 0.001, η = 0.85; Target Location Frequency: F(1,19) [F(1, 23) = 1.68, p = 0.20, η = 0.06]. However, for RT, significant = 10.92, p = 0.003, η = 0.36] and a significant interaction [F(1,19) main effects were found for Cue [F(1, 23) = 97.04, p < 0.0001, η = 10.15, p = 0.004, η = 0.34]. A post hoc t-test also confirmed the = 0.80] and Target Location Frequency [F(1, 23) = 8.27, p = 0.008, prevalence of top-down AC over SL, which could be seen only in η = 0.26], but the interaction between these factors was not signifi- the neutral cue trials (neutral: [t(19) = -3.41, p = 0.002, Cohen’s d = cant [F(1, 23) = 2.21, p = 0.15, η = 0.08]. In contrast to the main -0.43, -40 sec]; valid: [t(19) = -1.57, p = 0.13, Cohen’s d = -0.08, -7 analysis, the effect of Target Location Frequency was in the opposite ms]). However, for accuracy we only found a main effect of cue [Cue: direction, showing shorter RTs in the intermediate-low (vs. interme- F(1,19) = 10.63, p = 0.004, η = 0.35; Target Location Frequency: diate-high) condition. Since in the present study we wanted to assess F(1,19) = 1.23, p = 0.28, η = 0.06], and no significant interaction the top-down control in the presence of a clear SL effect, we did not [F(1,19) = 0.41, p = 0.52, η = 0.02]. consider the intermediate frequency locations further. 1 3 Attention, Perception, & Psychophysics 1 3 Attention, Perception, & Psychophysics ◂Fig. 3 Sensor plots showing contra (black line), ipsi (red line) and cue (t(23) = 0.76, p = 0.22, Cohen's d = 0.15), and not for the difference waves (contra-minus-ipsi; blue line) activity following targets in HFTL (following neutral cue: t(23) = −0.53, p = a neutral cue (a, b), or a valid cue (c, d). Panels a and c depict activ- 0.70, Cohen's d = −0.10; following valid cue: t(23) = 0.55, ity in the LFTL condition, and panels b and d depict activity in the p = 0.29, Cohen's d = 0.11). HFTL condition. Time-point zero indicates the search-display onset. The grey area is the time-window where mean amplitude of the P1 A two-way rm-ANOVA with Cue (valid, neutral) and Tar- was calculated, whereas the yellow area refers to the N2pc time- get Location Frequency (high, low) was performed to inves- range. Panels e and f show the mean amplitude of P1 (e) and N2pc tigate the effects of top-down AC and SL on the early stage (f), calculated by subtracting the contra-minus-ipsi channel, in the of target selection. Importantly, this was done on the contra- two Target Location Frequency conditions as a function of the cue. Error bars in plots e and f represent the standard errors of the mean minus-ipsi difference waves, hence characterizing lateraliza- tion effects. Similar to the CNV, this analysis revealed a sig- nificant main effect of Cue [F (1,23) = 23.20, p < 0.001, η Frequency [F(1,23) = 0.47, p = 0.497, η = 0.02]. Impor- = 0.50] that elicited a larger P1 lateralization for valid (vs. tantly, a significant interaction emerged between the two neutral) cues, but not of Target Location Frequency [F(1,23) 2 2 factors [F(1,23) = 4.28, p = 0.049, η = 0.15]. Post hoc = 0.65, p = 0.424, η = 0.02]. Furthermore, no significant p p paired t-tests (two-tailed) revealed that targets in the LFTL interaction emerged between the two factors [F(1,23) = 0.58, condition following a valid cue elicited a larger N2pc com- p = 0.451, η = 0.02] (Fig. 3). pared with targets at that location following a neutral cue [t(23) = 2.85, p = 0.009, Cohen’s d = 0.22; 0.74 μV]. In contrast, this effect was not present for targets in the HFTL Discussion condition [t(23) = -0.10, p = 0.918, Cohen’s d = -0.006; -0.02 μV]. Furthermore, no significant difference in N2pc In the current study, we aimed to assess the combined amplitudes was found between HFTL and LFTL, either in effects and neural correlates of top-down AC and SL, the neutral [t(23) = -0.94, p = 0.355, Cohen’s d = -0.37; when both are present. To that end, we manipulated top- -1.28 μV] or in the valid cue condition [t(23) = -0.40, p = down AC via endogenous cueing, and we introduced an 0.688, Cohen’s d = -0.16; -0.51 μV] (Fig. 3). imbalance of in-target frequency across locations in the same visual search task. Critically, we implemented the CNV target location imbalance only for neutrally cued trials in order to fully dissociate target location frequency To assess whether a valid endogenous cue elicited a pre- and cue validity in our task. Furthermore, we utilized a paratory effect, a one-way rm-ANOVA was performed neutral rather than an invalid cue as a baseline. Impor- with Cue (valid, neutral) on the CNV component. This tantly, to be able to compare and unify results regarding analysis showed a significant difference between the two the interaction between different AC mechanisms, and conditions [F(1,23) = 33.46, p < 0.001, η = 0.59]. Spe- to shed light on the functional architecture of visual cifically, a larger CNV was evoked by valid cues, indicating spatial attention, we used the same visual search task that the participants could prepare to orient their attentional (with some adjustments due to methodological reasons) resources before the search array onset following an inform- already implemented and adapted for the study of the ative cue. Furthermore, we explored whether SL proactively integrated effect of other AC signals, namely top-down modulates top-down control, such that a preparatory effect control via endogenous cueing and bottom-up allocation would emerge according to the target location frequencies. of attention due to salience (Beffara et al., 2022; Rashal To this end, another rm-ANOVA was conducted on the data et al., 2022). from trials following a valid cue in the HFTL and LFTL conditions. No significant difference was found between Combined effect of top‑down control and SL these two conditions [F(1,23) = 0.70, p = 0.41, η = 0.02] on behaviour and on N2pc (Fig. 4). The behavioural results concerning SL and top-down con- P1 trol confirmed our hypotheses and were in line with the literature, showing an overall effect of both mechanisms Similar to the analysis conducted for the N2pc, we per- of facilitation of target identification following valid, com- formed a one-sample t-test (one-tailed) to test whether P1 pared to neutral, cues (e.g., Folk et al., 1992; Posner, 1980; was meaningfully lateralized in each condition (mean ampli- Rashal et al., 2022), as well as targets presented in the high- tude greater than zero). Results showed a significant laterali- (vs. low-) frequency location (Ferrante et al., 2018; Geng zation for targets in the LFTL following a valid cue (t(23) = & Behrmann, 2005). Participants could indeed benefit from 2.30, p = 0.01, Cohen's d = 0.46), but not following a neutral the available information and prepare for the onset of the 1 3 Attention, Perception, & Psychophysics Fig. 4 The plot on the left shows the CNV (contingent negative varia- ing to the LFTL (dashed line) and HFTL (dotted line). Time-point tion) elicited by neutral (black line) and valid (red line) cues, whereas zero indicates cue onset. The yellow area represents the time-window the plot on the right shows the CNV elicited by the valid cue point- where the mean amplitude of the CNV was quantified array, and then efficiently identify the relevant item (i.e., tar - a gating mechanism in order to fully and efficiently guide get). Furthermore, participants’ performance indicated that attention to current objectives. they had learnt the bias induced by the statistical imbalance Previous studies, however, are more in line with the idea of target frequency across locations, which could facilitate that top-down control and SL are independent mechanisms target detection in the location where it was more likely and, when active together, the effects of the two are summed- to appear. During the debriefing at the end of the experi- up to bias the competition over attentional resources (Gao mental session, only four subjects reported having noticed & Theeuwes, 2020; Geng & Behrmann, 2005). Gao and the manipulation. The main results did not significantly Theeuwes (2020) argued that at the level of neural activ- change by excluding their data, supporting the idea that ity, statistical learning creates an implicit landscape where people can implicitly extract regularities from their exter- multiple spatial locations have a certain level of activations nal environment even without explicit instructions (Ferrante and inhibitions, and top-down control may then operate to et al., 2018; Fiser & Aslin, 2001; Jiang, 2018; Saffran et al., orient the attentional spotlight from one location to another 1996). (Gao & Theeuwes, 2020). Their theory seems to be in line Most importantly, in this study we directly examined with our EEG results. Indeed, even if in our study the final whether these two AC signals affect attentional deploy - attentional choice seems to be guided by top-down control ment in an independent manner when acting together, or that prevails over SL in determining performance (e.g., RTs), whether the effect of one signal interferes with the effect the EEG data showed that SL is not completely overridden of the other. Our results were more in line with the lat- by top-down AC, as an interaction is demonstrated by the ter hypothesis, revealing an interaction between the two N2pc modulation. sources of AC. In particular, at the behavioural level, we That is, in terms of the N2pc, the benefit of cueing the observed a gating effect of top-down control over SL, as target location was associated with a larger N2pc that was the effect of SL clearly emerged only in the absence of elicited by targets following a valid, compared to a neutral, top-down guidance; responses were faster for targets in the cue. In addition, this effect interacted with SL, showing an HFTL compared with the LFTL following a neutral but not increased N2pc amplitude for validly cued targets (com- a valid cue. The prevalence of top-down control could rep- pared with neutrally cued targets), but only when shown resent an important feature of the functional architecture at a low-frequency location. This appears to be in line with of visual spatial attention, and it could reflect the ability the priority map theory, whereby the low-frequency target of voluntary control to actively inhibit (or at least reduce) location in general should be associated with less neuronal the contribution of other signals, for example, SL, through activation due to the statistical learning mechanism, and 1 3 Attention, Perception, & Psychophysics thus can benefit from the allocation of top-down attentional across locations, while keeping a location in spatial work- resources, as guided by a valid cue. In contrast, when the ing memory in order to induce top-down control (Awh & target appeared at the high-frequency location, any potential Jonides, 2001; Munneke et al., 2010). Therefore, in their benefit from the preceding cue was abolished. This might experiment, only SL impacted the search strategy, while suggest that the gating effect exerted by top-down AC over the cue determined the second response. Furthermore, in SL that we observed on behaviour was not a full gating; on the present study, there was no invalid cue condition and the contrary, at least at some stage along the target selec- the informative cue predicted the target location with 100% tion process the SL was still exerting an effect that could be validity. Thus, participants could fully trust the informa- strong enough to prevent top-down control from emerging tion coming from the top-down control mechanism. There- (as observed for targets at the high-frequency location in the fore, when these AC mechanisms act together, the inter- current study). action between SL and top-down control may depend on Interestingly, the different pattern of N2pc amplitudes the degree of validity of the latter, in turn determining the observed in the two location-frequency conditions could strength with which it can guide attentional selection even suggest that a serial search strategy was employed by the to the point of bypassing the information coming from SL. participants to find the target (e.g., Woodman & Luck, 2003). Together, our findings suggest a close and complex inter - That is, the participants may have first searched the high-fre- action between top-down control and SL, where when one quency location before moving to the intermediate- and low- mechanism is acting, and is potentially strong enough to frequency locations. The positive peak observed in the low- optimize selection, the effect of the other is reduced. frequency condition at the N2pc time-window supports this However, we have to note that although the present view (Fig. 3a), considering that the high- and low-frequency experiment contained a number of trials that should be suf- locations were positioned at exactly opposite from each other ficient to detect significant effects on the EEG markers of on the screen. Still, this positive peak does not seem to be a interest with our sample size (Boudewyn et al., 2018), it fully inverted N2pc, thus, at this point this is only a specula- seems to be limited in its power to capture smaller effects tion that could benefit from further investigation. we observe as significant (e.g. Ngiam et al., 2021). Given One could argue that the size of the N2pc was reduced that a specific target-frequency condition was associated due to the target appearing at the same location repeatedly with just one hemisphere depending on the subject group across consecutive trials, as there was less need for the allo- (e.g., HFTL in the right hemisphere), in the current study cation of attentional resources if attention was already at the the N2pc was calculated by collapsing the left and right correct location. This seems consistent with the cueing effect hemispheres only at a group level (see Wang et al., 2019, for we observed, since the SL manipulation was present only in a similar experimental paradigm). This might have made the the neutrally cued trials. That is, the target was more likely data vulnerable to increased noise from residual inter-indi- to appear repeatedly at the same location in the neutral cue vidual variability related to asymmetries in brain activity. condition than in the valid cue condition. In contrast, in the This, in turn, might have caused reduced statistical power in valid cue condition there was less chance of repeating target some conditions. Therefore, in future studies, larger sample location across trials, as target location was equally probable sizes and number of trials might be adopted to increase the in that condition. In this vein, the decrease in mean N2pc robustness of SL and cueing effects studied here. amplitude in the neutral cue trials is in line with the find- ings of van Moorselaar and Slagter (2019), who showed a Neural activity underlying the preparatory effect reduction in N2pc amplitude for targets presented at the end of top‑down control of a sequence of targets presented at one location, compared to the targets presented at the beginning of the sequence. In this study, the CNV results reflected processes involved However, this alternative explanation cannot account for in the preparation of anticipatory attention for the upcoming the SL effect we observed, as it predicts a larger N2pc in stimulus and motor preparation needed to respond (Brunia & the LFTL compared to the HFLT, the latter having more van Boxtel, 2001; Tecce, 1972). It has been shown that the chances of repeating target location in consecutive trials than CNV mean amplitude was modulated by the presentation of the former. Evidently, this is the opposite of the pattern we a warning stimulus, such as a cue, demonstrating its link to observed here. strong attentional engagement (e.g., Rashal et al., 2022; Sch- One of the main differences between this study and the evernels et al., 2014; van den Berg et al., 2014). During the earlier ones (e.g., Gao & Theeuwes, 2020) is that in our CTI, a larger CNV was indeed elicited by a valid (vs. neu- paradigm both top-down control and SL directly determine tral) cue, in accordance with the idea that under top-down the search strategy, making it possible to investigate how control participants could orient their attention in advance much SL leaks onto cued trials. In contrast, Gao and Theeu- toward a certain position. In addition, as a consequence of wes (2020) manipulated the frequency of target occurrence the preparatory effect exerted by top-down control, we found 1 3 Attention, Perception, & Psychophysics that the valid cue could also affect the early components of general gating effect of top-down control, but might simply target selection, such as the P1 (e.g., Mangun & Hillyard, index that the SL became a condition-dependent mechanism, 1991). Knowing the upcoming target location allows an in this case, following the neutral cue. However, this hypoth- early allocation of attentional resources toward a specific esis can be excluded given the presence of an interaction region of that display. Indeed, targets following a valid cue between top-down control and SL on the N2pc, which sug- produced a general contralateral enhancement of P1 ampli- gests an intervention of SL in assigning different attentional tudes. Importantly, we did not find any preparatory advan- priorities to different locations on the priority map, which tage due to statistical regularities for the two components; can lead to a reduction of the benefit of top-down control no difference emerged between the CNV elicited by valid during the early stage of target selection. cues pointing to the high-frequency location and valid cues One critical aspect is that the ability of top-down con- pointing to the low-frequency location, and the P1 to the trol to modulate attention and bypass the information of target was not modulated by the target location. all the other AC mechanisms can depend on its strength and its relevance in the given experimental context. The priority map is modulated by combined effects Indeed, a fully reliable informative cue, as the one used in of different attention control signals this experiment, can strongly guide attention toward the instructed location without additional information from Most studies of attention control, however, use experimental other AC signals. Contrary to that, when the informative paradigms that address only one specific AC mechanism at a cue is partially predictive (such as is the case where some time, which makes it difficult to understand the contribution invalid cue trials occurred), the gating effect of top-down of each attentional signal in generating the final attentional control is weaker since it is needs to also consider infor- choice, and controls how goal-directed behaviour is accom- mation coming from other AC signals, in this case, SL. plished by the brain. Here we used a visual search paradigm previously imple- mented in its general form in another study that examined the combined effect of different attention-control sources Conclusion with behaviour and EEG, in an attempt to develop a uni- fied account. Rashal and colleagues (2022) demonstrated This study seems to indicate an interaction between top-down that top-down guidance of attention via an endogenous cue control and SL, where, when one mechanism is at a play, the diminished the benefit of target salience and the interference influence of the other is reduced or even abolished. In particu- from a salient distractor. Similarly, top-down control seems to lar, the collected behavioural results suggest that when strong prevail over the other AC signal, i.e., SL, in this experiment. top-down attention control is available, SL could not emerge, However, our EEG data showed that SL can block the benefit as if the information coming from the cue and guiding atten- of a valid cue in the high-frequency location, if in that loca- tion to the indicated spatial location bypasses the information tion the neural activity already reached the highest possible coming from the implicit learning process. Indeed, the fully peak due to the probability distribution of target frequency. reliable valid cue allowed participants to pre-allocate their As mentioned above, since the present work and the work attentional resources to the upcoming target location before of Rashal and colleagues (2022) used the same experimental the stimuli array onset, thus fully optimizing subsequent tar- task, it is possible to formulate hypotheses on the functional get selection. Nevertheless, our EEG results suggest that SL architecture of visual spatial attention. In particular, together was not totally overridden, at least at some stages of target these findings seem to suggest that there is a general domi- selection, and in turn being able to reduce the impact of top- nance of the mechanism underlying top-down allocation down guidance in some cases. of attention, since the effects of both bottom-up capture by Author note This study is part of a collaborative project (MAC-Brain: salience and SL on performance were diminished, or even Developing a Multi-scale account of Attentional Control as the con- absent, following a valid cue. In this scenario, one could straining interface between vision and action: A cross-species inves- argue that the final attentional choice is the result of the tigation of relevant neural circuits in the human and macaque Brain") funded under the European FLAG-ERA JTC 2017 program and associ- activity of only one mechanism, i.e., top-down control, that ated with the Human Brain Project. prevents the influence of all the others. Still, since in our experiment the cue was 100% valid and predicted the tar- Funding Open access funding provided by Università degli Studi di get location well in advance, participants could ignore the Verona within the CRUI-CARE Agreement. information coming from the target probability distribution Open practices statement Data will be made available on EBRAINS. when they already had certain and explicit information about Other materials for the experiments reported here are available upon where to find the target. As a consequence, the lack of effect request. None of the experiments was preregistered. of SL in the valid cue condition was likely not the result of a 1 3 Attention, Perception, & Psychophysics Carrasco, M. (2011). Visual attention: The past 25 years. Vision Declarations Research, 51(13), 1484–1525. https:// doi. org/ 10. 1016/j. visres. 2011. 04. 012 Conflict of interest The authors have no conflicts of interest to declare. Chelazzi, L., & Santandrea, E. (2018). The time constant of atten- tional control: Short, Medium and Long (Infinite?). Journal of Open Access This article is licensed under a Creative Commons Attri- Cognition, 1(1). Ubiquity Press. https://doi. or g/10. 5334/ joc. 24 bution 4.0 International License, which permits use, sharing, adapta- Chelazzi, L., Perlato, A., Santandrea, E., & Della Libera, C. (2013). tion, distribution and reproduction in any medium or format, as long Rewards teach visual selective attention. Vision Research, 85, as you give appropriate credit to the original author(s) and the source, 58–72. https:// doi. org/ 10. 1016/j. visres. 2012. 12. 005 provide a link to the Creative Commons licence, and indicate if changes Chelazzi, L., Eštočinová, J., Calletti, R., Gerfo, E. . Lo., Sani, I., were made. The images or other third party material in this article are Libera, C. Della., & Santandrea, E. (2014). Altering spatial included in the article's Creative Commons licence, unless indicated priority maps via reward-based learning. Journal of Neurosci- otherwise in a credit line to the material. If material is not included in ence, 34(25), 8594–8604. https://do i.o rg/1 0.15 23/J NEURO SCI. the article's Creative Commons licence and your intended use is not 0277- 14. 2014 permitted by statutory regulation or exceeds the permitted use, you will Desimone, R., & Duncan, J. (1995). Neural mechanisms of selective need to obtain permission directly from the copyright holder. To view a visual attention. Annual Review of Neuroscience, 18, 193–222. copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . Annual Reviews Inc. https:// doi. or g/ 10. 1146/ annur e v . ne. 18. 030195. 001205 Di Bello, F., Ben HadjHassen, S., Astrand, E., & Ben Hamed, S. References (2022). Prefrontal control of proactive and reactive mechanisms of visual suppression. Cerebral cortex (New York, N.Y. : 1991), 32(13), 2745–2761. https:// doi. org/ 10. 1093/ cercor/ bhab3 78 Aslin, R. N., & Newport, E. L. (2012). Statistical learning: From Druker, M., & Anderson, B. (2010). Spatial probability aids visual acquiring specific items to forming general rules. Current stimulus discrimination. Frontiers in Human Neuroscience, 4. Directions in Psychological Science, 21(3), 170–176. https:// https:// doi. org/ 10. 3389/ fnhum. 2010. 00063 doi. org/ 10. 1177/ 09637 21412 436806 Duncan, D., & Theeuwes, J. (2020). Statistical learning in the Awh, E., & Jonides, J. (2001). Overlapping mechanisms of attention absence of explicit top-down attention. Cortex, 131, 54–65. and spatial working memory. Trends in Cognitive Sciences, 5(3), https:// doi. org/ 10. 1016/j. cortex. 2020. 07. 006 119–126. Egeth, H. E., & Yantis, S. (1997). VISUAL ATTENTION: Control, Awh, E., Belopolsky, A. V., & Theeuwes, J. (2012). Top-down versus Representation, and Time Course. Annual Review of Psychol- bottom-up attentional control: A failed theoretical dichotomy. ogy, 48. www. annua lrevi ews. org Trends in Cognitive Sciences, 16(8), 437–443. https:// doi. org/ Eimer, M. (1994). An ERP study on visual spatial priming with 10. 1016/j. tics. 2012. 06. 010 peripheral onsets. Psychophysiology, 31(2), 154–163. https:// Bacigalupo, F., & Luck, S. J. (2019). Lateralized suppression of doi. org/ 10. 1111/j. 1469- 8986. 1994. tb010 35.x alpha-band EEG activity as a mechanism of target processing. Eimer, M. (1996). The N2pc component as an indicator of attentional Journal of Neuroscience, 39(5), 900–917. https:// doi. org/ 10. selectivity. Electroencephalography and clinical Neurophysiol- 1523/ JNEUR OSCI. 0183- 18. 2018 ogy, 99, 225–234. Baumgartner, H. M., Graulty, C. J., Hillyard, S. A., & Pitts, M. A. Failing, M., & Theeuwes, J. (2018). Selection history: How reward (2018). Does spatial attention modulate the earliest component modulates selectivity of visual attention. Psychonomic Bulletin of the visual evoked potential? Cognitive Neuroscience, 9(1–2), and Review, 25(2), 514–538. Springer New York LLC. https:// 4–19. https:// doi. org/ 10. 1080/ 17588 928. 2017. 13334 90 doi. org/ 10. 3758/ s13423- 017- 1380-y Beffara, B., Hadj-Bouziane, F., Hamed, S. B., Boehler, C. N., Ferrante, O., Patacca, A., Di Caro, V., Della Libera, C., Santandrea, Chelazzi, L., Santandrea, E. & Macaluso, E. (2022). Dynamic E., & Chelazzi, L. (2018). Altering spatial priority maps via sta- causal interactions between occipital and parietal cortex explain tistical learning of target selection and distractor filtering. Cor - how endogenous spatial attention and stimulus-driven salience tex, 102, 67–95. https:// doi. org/ 10. 1016/j. cortex. 2017. 09. 027 jointly shape the distribution of processing priorities in 2D vis- Fiser, J., & Aslin, R. N. (2001). Unsupervised statistical learning of ual space. Neuroimage, 255, 119206. https://doi. or g/10. 1016/j. higher-order spatial structures from visual scenes. Psychologi- neuro image. 119206 cal Science, 12(6), 499–504. Ben-Shachar, M. S., Lüdecke, D., & Makowski, D. (2020). effectsize: Folk, C. L., Remington, R. W., & Johnston, J. C. (1992). Involuntary Estimation of effect size indices and standardized parameters. covert orienting is contingent on attentional control settings. Journal of Open Source Software, 5(56), 2815. https:// doi. org/ Journal of Experimental Psychology. Human Perception and 10. 21105/ joss. 02815 Performance, 18(4), 1030–1044. https://doi. or g/10. 1037/ 0096- Bisley, J. W. (2011). The neural basis of visual attention. Journal 1523. 18.4. 1030 of Physiology, 589(1), 49–57. https://doi. or g/10. 1113/ jph ysiol. Gao, Y., & Theeuwes, J. (2020). Independent effects of statistical 2010. 192666 learning and top-down attention. Attention, Perception, and Bisley, J. W., & Goldberg, M. E. (2010). Attention, intention, and Psychophysics, 82(8), 3895–3906. https:// doi. or g/ 10. 3758/ priority in the parietal lobe. Annual Review of Neuroscience, 33, s13414- 020- 02115-x 1–21. https:// doi. org/ 10. 1146/ annur ev- neuro- 060909- 152823 Geng, J. J., & Behrmann, M. (2002). Probability cuing of target loca- Boudewyn, M. A., Luck, S. J., Farrens, J. L., & Kappenman, E. S. tion facilitates visual search implicitly in normal participants (2018). How many trials does it take to get a significant ERP and patients with hemispatial neglect. Psychological Science, effect? It depends. Psychophysiology, 55(6), e13049. https://d oi. 16, 520–525. org/ 10. 1111/ psyp. 13049 Geng, J. J., & Behrmann, M. (2005). Spatial probability as atten- Brunia, C. H. M., & van Boxtel, G. J. M. (2001). Wait and see. Inter- tional cue. Perception & Psychophysics, 67(7), 1252–1268. national Journal of Psychophysiology, 43(1), 59–75. https://d oi. https://link .spr ing er.com/ conte nt/ pdf/ 10. 3758/ BF031 93557. pdf org/ 10. 1016/ S0167- 8760(01) 00179-9 1 3 Attention, Perception, & Psychophysics Grent-‘t-Jong, T., & Woldorff, M. G. (2007). Timing and sequence of Newport, E. L., Hauser, M. D., Spaepen, G., & Aslin, R. N. (2004). brain activity in top-down control of visual-spatial attention. PLoS Learning at a distance: II. Statistical learning of non-adjacent Biology, 5(1), 12. https:// doi. org/ 10. 1371/ journ al. pbio. 00500 12 dependencies in a non-human primate. Cognitive Psychology, Heinze, H. J., Mangun, G. R., Burchert, W., Hinrichs, H., Scholz, M., 49(2), 85–117. Münte, T. F., Gös, A., Scherg, M., Johannes, S., Hundeshagen, Ngiam, W. X. Q., Adam, K. C. S., Quirk, C., Vogel, E. K., & Awh, E. H., Gazzaniga, M. S., & Hillyard, S. A. (1994). Combined spatial (2021). Estimating the statistical power to detect set size effects in and temporal imaging of brain activity during visual selective contralateral delay activity. Psychophysiology, 58, e13791. https:// attention in humans. Nature, 372(6506), 543–546. https://doi. or g/ doi. org/ 10. 1111/ psyp. 13791 10. 1038/ 37254 3a0 Noudoost, B., Chang, M. H., Steinmetz, N. A., & Moore, T. (2010). Ipata, A. E., Gee, A. L., Bisley, J. W., & Goldberg, M. E. (2009). Neu- Top-down control of visual attention. In Current Opinion in Neu- rons in the lateral intraparietal area create a priority map by the robiology, 20(2), 183–190. Elsevier Ltd. https://doi. or g/10. 1016/j. combination of disparate signals. Experimental Brain Research, conb. 2010. 02. 003 192(3), 479–488. https:// doi. org/ 10. 1007/ s00221- 008- 1557-8 Parisi, G., Mazzi, C., Colombari, E., Chiarelli, A. M., Metzger, B. A., Jiang, Y. V. (2018). Habitual versus goal-driven attention. Cortex, 102, Marzi, C. A., & Savazzi, S. (2020). Spatiotemporal dynamics of 107–120. Masson SpA. https://doi. or g/10. 1016/j. cor tex.2017. 06. attentional orienting and reorienting revealed by fast optical imag- 018 ing in occipital and parietal cortices. Neuroimage, 222, 117244. Kiss, M., Van Velzen, J., & Eimer, M. (2008). The N2pc component https:// doi. org/ 10. 1016/j. neuro image. 2020. 117244 and its links to attention shifts and spatially selective visual pro- Peirce, J. W. (2007). PsychoPy-Psychophysics software in python. cessing. Psychophysiology, 45(2), 240–249. https:// doi. org/ 10. Journal of Neuroscience Methods, 162(1–2), 8–13. https:// doi. 1111/j. 1469- 8986. 2007. 00611.xorg/ 10. 1016/j. jneum eth. 2006. 11. 017 Lawrence, M. (2015). ez: Easy analysis and visualization of factorial Posner, M. I. (1980). Orienting of attention. The Quarterly Journal of experiments [R package version 4.3.]. https:// github. com/ mike- Experimental Psychology, 32(1), 3–25. https:// doi. org/ 10. 1080/ lawre nce/ ez (Original work published 2011)00335 55800 82482 31 Leber, A. B., & Egeth, H. E. (2006). It’s under control: Top-down Ptak, R. (2012). The frontoparietal attention network of the human search strategies can override attentional capture. Psychonomic brain: Action, saliency, and a priority map of the environment. Bulletin & Review, 13(1), 132–138. https://d oi.o rg/1 0.3 758/b f031 Neuroscientist, 18(5), 502–515. h tt ps :/ / d oi . o rg/ 1 0 . 1 17 7/ 1 0 73 8 9382458411 409051 Liebrand, M., Pein, I., Tzvi, E., & Krämer, U. M. (2017). Tempo- R Core Team. (2016). R: A language and environment for statistical ral dynamics of proactive and reactive motor inhibition. Fron- computing. https:// www. gbif. org/ tool/ 81287/r- a- langu age- and- tiers in Human Neuroscience, 11, 204. https://doi. or g/10. 3389/ envir onment- for- stati stical- compu ting fnhum. 2017. 00204 Rashal, E., Senoussi, M., Santandrea, E., Ben-Hamed, S., Macaluso, Liesefeld, H. R., Liesefeld, A. M., Töllner, T., & Müller, H. J. (2017). E., Chelazzi, L., & Boehler, C. N. (2022). An EEG study of the Attentional capture in visual search: Capture and post-capture combined effects of top-down and bottom-up attentional selection dynamics revealed by EEG. NeuroImage, 156, 166–173. https:// under varying task difficulty. Psychophysiology. https:// doi. org/ doi. org/ 10. 1016/j. neuro image. 2017. 05. 01610. 1111/ psyp. 14002 Livingstone, A. C., Christie, G. J., Wright, R. D., & McDonald, J. Reynolds, J. H., & Chelazzi, L. (2004). Attentional modulation of J. (2017). Signal enhancement, not active suppression, follows visual processing. Annual Review of Neuroscience, 27, 611–647. the contingent capture of visual attention. Journal of Experi-https:// doi. org/ 10. 1146/ annur ev. neuro. 26. 041002. 131039 mental Psychology: Human Perception and Performance, 43(2), Reynolds, J. H., & Heeger, D. J. (2009). The normalization model 219–224. of attention. Neuron, 61(2), 168–185. https:// doi. org/ 10. 1016/j. Luck, S. J., & Hillyard, S. A. (1994). Spatial filtering during visual neuron. 2009. 01. 002 search: Evidence from human electrophysiology. Journal of Rosa-Salva, O., Fiser, J., Versace, E., Dolci, C., Chehaimi, S., Santo- Experimental Psychology: Human Perception and Performance, lin, C., & Vallortigara, G. (2018). Spontaneous learning of visual 20(5), 1000–1014. structures in domestic chicks. Animals, 8(8), 135. https://doi. or g/ Luck, S. J., Girelli, M., Mcdermott, M. T., & Ford, M. A. (1997). 10. 3390/ ani80 80135 Bridging the gap between monkey neurophysiology and human Saffran, J. R., Aslin, R. N., & Newport, E. L. (1996). Statistical learn- perception: An ambiguity resolution theory of visual selective ing by 8-month-old infants. Cognition, 274, 1–10. attention. Cognitive Psychology, 33, 64–87. Santolin, C., Rosa-Salva, O., Vallortigara, G., & Regolin, L. (2016). Luck, S. J., Woodman, G. F., & Vogel, E. K. (2000). Event-related Unsupervised statistical learning in newly hatched chicks. Cur- potential studies of attention. Trends in Cognitive Sciences, rent Biology, 26(23), R1218–R1220. https:// doi. org/ 10. 1016/j. 4(11), 432–440. h t t p s : / / d o i . o r g / 1 0 . 1 0 1 6 / S 1 3 6 4 - 6 6 1 3 ( 0 0 ) cub. 2016. 10. 011 01545-X Schevernels, H., Krebs, R. M., Santens, P., Woldorff, M. G., & Boe- Macaluso, E., & Doricchi, F. (2013). Attention and predictions: hler, C. N. (2014). Task preparation processes related to reward Control of spatial attention beyond the endogenous-exogenous prediction precede those related to task-difficulty expectation. dichotomy. Frontiers in Human Neuroscience (OCT). Frontiers NeuroImage, 84, 639–647. https://doi. or g/10. 1016/j. neur oimag e. Media S. A. https:// doi. org/ 10. 3389/ fnhum. 2013. 006852013. 09. 039 Mangun, G. (1995). Neural mechanisms of visual selective attention. Serences, J. T., & Yantis, S. (2007). Spatially selective representations Psychophysiology, 32, 4e18. of voluntary and stimulus-driven attentional priority in human Mangun, G. R., & Hillyard, S. A. (1991). Modulations of sensorye- occipital, parietal, and frontal cortex. Cerebral Cortex, 17(2), voked brain potentials indicate changes in perceptual processing 284–293. https:// doi. org/ 10. 1093/ cercor/ bhj146 during visual-spatial priming. Journal Experimental Psychology Tecce, J. J. (1972). Contingent negative variation (CNV) and psycho- Human, 17, 1057e1074. logical processes in man. Psychological Bulletin, 77(2), 73–108. Munneke, J., Heslenfeld, D. J., & Theeuwes, J. (2010). Spatial work-https:// doi. org/ 10. 1037/ h0032 177 ing memory effects in early visual cortex. Brain and Cognition, Theeuwes, J. (1991). Exogenous and endogenous control of attention: 72(3), 368–377. The effect of visual onsets and offsets. Perception & Psychophys- ics, 49(1), 83–90. 1 3 Attention, Perception, & Psychophysics Theeuwes, J. (1994). Stimulus-driven capture and attentional set: Walter, W. G., Cooper, R., Aldridge, V. J., McCallum, W. C., & Win- Selective search for color and visual abrupt onsets. Journal of ter, A. L. (1964). Contingent negative variation: An electric sign Experimental Psychology: Human Perception and Performance, of sensorimotor association and expectancy in the human brain. 20(4), 799–806. https:// doi. org/ 10. 1037/ 0096- 1523. 20.4. 799 Nature., 203(4943), 380–384. https:// doi. org/ 10. 1038/ 20338 0a0 Theeuwes, J. (2010). Top-down and bottom-up control of visual selec- Wang, B., van Driel, J., Ort, E. & Theeuwes, J. (2019). Anticipatory tion. Acta Psychologica, 135(2), 77–99. https://doi. or g/10. 1016/j. distractor suppression elicited by statistical regularities in visual actpsy. 2010. 02. 006 search. Journal of Cognitive Neuroscience, 1–14. https:// doi. org/ Theeuwes, J., & Godijn, R. (2004). Inhibition-of-return and oculomotor 10. 1162/ jocn_a_ 01433 interference. Vision Research, 44(12), 1485–1492. https://doi. or g/ Woodman, G. F., & Luck, S. J. (2003). Serial deployment of attention 10. 1016/j. visres. 2003. 09. 035 during visual search. Journal of Experimental Psychology. Human Van Den Berg, B., Krebs, R. M., Lorist, M. M., & Woldorff, M. G. Perception and Performance, 29(1), 121–138. https:// doi. org/ 10. (2014). Utilization of reward-prospect enhances preparatory 1037// 0096- 1523. 29.1. 121 attention and reduces stimulus conflict. Cognitive, Affective and Wu, C.-T., Libertus, M. E., Meyerhoff, K. L., & Woldorff, M. G. Behavioral Neuroscience, 14(2), 561–577. https:// doi. or g/ 10. (2011). The temporal dynamics of object processing in visual 3758/ s13415- 014- 0281-z cortex during the transition from distributed to focused spatial Van Moorselaar, D., & Slagter, H. A. (2019). Learning what is irrel- attention. J. Cognitive Neuroscience, 23, 4094–4105. evant or relevant: Expectations facilitate distractor inhibition and Yantis, S., & Egeth, H. E. (1999). On the distinction between vis- target facilitation through distinct neural mechanisms. Journal of ual salience and stimulus-driven attentional capture. Journal of Neuroscience, 39(35), 6953–6967. https:// doi. or g/ 10. 1523/ JNEUR Experimental Psychology: Human Perception and Performance, OSCI. 0593- 19. 2019 25(3), 661–676. https:// doi. org/ 10. 1037/ 0096- 1523. 25.3. 661 Van Voorhis, S., & Hillyard, S. A. (1977). Visual evoked potentials and Zivony, A., Allon, A. S., Luria, R., & Lamy, D. (2018). Dissociating selective attention to points in space. Perception & Psychophysics, between the N2pc and attentional shifting: An attentional blink 22(1), 54–62. https:// doi. org/ 10. 3758/ bf032 06080 study. Neuropsychologia, 121, 153–163. https://doi. or g/10. 1016/j. Verleger, R., Wascher, E., Wauschkuhn, B., Jaśkowski, P., Allouni, B., neuro psych ologia. 2018. 11. 003 Trillenberg, P., & Wessel, K. (1999). Consequences of altered cer- ebellar input for the cortical regulation of motor coordination, as Publisher’s note Springer Nature remains neutral with regard to reflected in EEG potentials. Experimental brain research, 127(4), jurisdictional claims in published maps and institutional affiliations. 409–422. https:// doi. org/ 10. 1007/ s0022 10050 809 1 3

Journal

Attention Perception & PsychophysicsSpringer Journals

Published: Aug 1, 2023

Keywords: N2pc; P1; Statistical learning; Endogenous cueing; Attention control; Priority map

There are no references for this article.