Spatial distribution of attentional bias in visuo-spatial working memory following multiple cues.

Acta Psychologica 150 (2014) 1–13

Contents lists available at ScienceDirect

Acta Psychologica journal homepage: www.elsevier.com/ locate/actpsy

Spatial distribution of attentional bias in visuo-spatial working memory following multiple cues Fabiano Botta ⁎, Juan Lupiáñez Department of Experimental Psychology, Mind, Brain, and Behavior Research Center, University of Granada, Spain

a r t i c l e

i n f o

Article history: Received 17 October 2013 Received in revised form 17 March 2014 Accepted 29 March 2014 Available online xxxx PsycINFO classification: 2346 Attention Keywords: Visuo-spatial working memory Spatial attention Attentional split Endogenous attention Exogenous attention

a b s t r a c t When attention is focused on one location, its spatial distribution depends on many factors, such as the distance between the attended location and the target location, the presence of visual meridians in between them, and the way, endogenous or exogenous, by which attention is oriented. However, it is not well known how attention distributes when more than one location is endogenously or exogenously cued, which was the focus of the current study. Furthermore, the distribution of attention has been manly investigated in perception. In the present study we faced this issue from a different perspective, by examining the spatial distribution of the attentional bias in visuo-spatial working memory (VSWM), when attention is oriented either exogenously or endogenously, i.e., after two peripheral vs. central symbolic cues (also manipulating cue–target predictability). Results indicated a systematic difference between endogenous and exogenous attention regarding the distribution of the attentional bias over VSWM. In fact, attentional bias following endogenous cues was affected by the presence of visual meridians and by the split of the attentional focus, converging in a unipolar attentional distribution, independently of cue–target predictability. On the other hand, when pulled by exogenous cues, attention distributed uni-modally or multi-modally depending on the distance between the cued locations, with larger effects for highly predictive cues. Results are discussed in terms of space-based, object-based and perceptual grouping mechanisms. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Spatial attention is important to select the most relevant inputs coming from the environment and process them faster and more accurately than irrelevant ones. It is broadly accepted that the control for attentional resources could be exerted, at least, in two different ways (e.g., Klein & Shore, 2000): endogenously and exogenously. Exogenous attention is driven by immediate physical properties of the cue, so by salient stimulation outside the observer, as when sudden changes in the environment attract both oculomotor responses and visual attention. On the contrary, endogenous control originates within the observer and requires development of a spatial expectancy on the basis of an intention usually developed accordingly to the predictability of a to be interpreted central symbolic cue. Most of our knowledge about endogenous and exogenous orienting comes from studies based on the typical Posner's (1980) cost and benefits paradigm. The main finding generally observed using this task is what is known as “facilitation” or “cuing effect”, in which stimuli presented at exogenously

⁎ Corresponding author at: Department of Experimental Psychology, University of Granada, Campus de Cartuja s/n 18011, 18011 Granada, Spain. Tel.: +34 958243763. E-mail address: [email protected] (F. Botta).

http://dx.doi.org/10.1016/j.actpsy.2014.03.013 0001-6918/© 2014 Elsevier B.V. All rights reserved.

cued or endogenously indicated locations are responded to faster and/or more accurately than those presented at uncued locations (e.g., Posner, 1980; Prinzmetal, Presti, & Posner, 1986). Even though cuing effects can be observed with both endogenous and exogenous cues, recent findings converged to the idea that endogenous attention and exogenous attention produce qualitatively different effects on information processing. Interesting examples of dissociation between them have been shown on information processing speed (Carrasco & McElree, 2001; Shore, Spence & Klein, 2001; Schneider & Bavelier, 2003), illusory perceptions (Chica, Charras, & Lupiáñez, 2008), conscious perception (Chica, Botta, Lupiáñez, & Bartolomeo, 2012; Chica et al., 2011), and on conflict resolution (Funes, Lupiáñez, & Milliken, 2007). Particularly, Funes et al. (2007), after showing good evidence for a double dissociation between endogenous and exogenous attentional mechanisms, suggested that a good strategy to investigate the differences between these ways of attentional orienting might be to take into account their differential effects on later stages of information processing. Among these stages, attentional modulation over working memory certainly represents an excellent candidate. As a matter of fact, in the last fifteen years strong relationships have been both theorized and observed between spatial attention and the encoding, maintenance, and retrieval of selected information in spatial

2

F. Botta, J. Lupiáñez / Acta Psychologica 150 (2014) 1–13

working memory (Awh & Jonides, 2001; Awh, Vogel, & Oh, 2006; Bundesen, 1990). Particularly, it has been proposed that attention might act as a “gatekeeper” for information storage in working memory, by controlling the flow of information into working memory, thus biasing the encoding of the objects that are most relevant at the present moment (see Awh et al., 2006, for a review). Botta, Santangelo, Raffone, Lupianez, and Olivetti Belardinelli (2010) explicitly investigated how exogenous orienting and endogenous orienting of spatial attention bias the information encoding into VSWM. Specifically, the presentation of an exogenous (Experiment 1) or endogenous (Experiment 2) cue was followed by a memory array consisting of 8 colored squares, one at each of eight possible locations evenly spaced around an imaginary circle. After a brief delay, the display was presented again with a single probe square in one of the locations, and participants were required to discriminate whether the color of the square presented at that location was the same or different from the one in the preceding memory array (see Luck & Vogel, 1997). The results pointed to a dissociation between the two orienting mechanisms in terms of both meridian and distance effects.1 Particularly, they found meridian crossing effects only when attention was oriented by endogenous (i.e., central symbolic and predictive) cues, in perfect accordance with the dissociation regarding the meridian effect already observed in the perceptual domain (Rizzolatti, Riggio, Dascola, & Umiltá, 1987; Rizzolatti, Riggio, & Sheliga, 1994). On the other hand, the bias exerted by exogenous (peripheral non predictive) cues on VSWM performance was purely affected by the cue–target distance. Summarizing, perceptual and VSWM studies indicated that when selective spatial attention is directed to a specific location it produces both an improvement in perceptual processing and an increase in the likelihood that information at this location will be encoded in VSWM. Notwithstanding, it seems that the spatial distribution of the attentional effects on both perception and VSWM changes depending on the way, endogenous or exogenous, by which attentional resources are allocated in the environment. In the present study we aimed at further exploring the nature of exogenous vs. endogenous attentional modulation over VSWM, by investigating the nature of the distribution of attention, and its modulation over VSWM, when attention is endogenously pushed toward to vs. exogenously pulled by more than one location/object at the same time, i.e., when two cues are simultaneously presented instead of one. In most cases the objects that capture our attention occupy one single and undivided spatial area. In these cases spatial distribution of attentional effects can be mostly explained by a simple gradient function of attentional facilitation, characterized by a peak over the attended location and by a decrease in the size of the effect as spatial distance from the attended location increases (Henderson & Macquistan, 1993). Consistently, classical models of spatial attention, as the “spotlight” metaphor originally proposed by Posner (1980) and the “zoom lens” model proposed by Eriksen and St. James (1986), both support an interpretation of spatial attention as unitary in nature, excluding the possibility that the attentional focus could be split in two or more disjointed locations. Notwithstanding, in many daily life circumstances, we need to select together stimuli that are located in noncontiguous regions of the space, as for objects partially occluded or when the to-be-selected stimulus consists of a configuration of objects at separated locations (see Bichot, Cave, & Pashler, 1999). In these cases selection might be better accomplished by other perceptual features than location, such as color or form. However, many studies suggest that selection mechanisms are mediated by spatial location even when the target stimulus

1 Meridian effect consists on an increase of RTs and/or errors when spatial cues and targets are presented on different visual fields (in reference to the vertical and/or horizontal meridians which represent the axes of symmetry of the visual field). Distance effect can be defined as a decrement in performance as a function of the spatial distance between the cue and the target.

is defined by other dimensions than position (see Cave & Bichot, 1999, for a review). This implies that location plays a crucial role in visual selection and somehow indicates that there may be a mechanism by which spatial attention selects noncontiguous locations. Accordingly, in the last twenty years there has been a growing body of evidence claiming that multiple locations can be simultaneously attended (Awh & Pashler, 2000; Baldauf & Deubel, 2008; Bichot et al., 1999; Carlson, VanRullen, Hogendoorn, Verstraten, & Cavanagh, 2007; Gobell, Tseng, & Sperling, 2004; Kraft et al., 2005; Müller, Malinowski, Gruber, & Hillyard, 2003; Scharlau, 2004). Therefore, at present the question regarding whether attention can be split or not is still unsolved, the reason perhaps being that the truth might lie in the middle, or that it might depend on the nature, endogenous or exogenous, of attentional orienting. In a very recent work, Feng and Spence (2012) intriguingly observed that since both the unitary and multiple-foci models are well supported by empirical and physiological data, it seems unlikely that one of these models is right and the other is wrong. They suggested instead that the occurrence of a single focus or multiple foci is possibly the result of specific experimental conditions. More specifically, Standage, Trappenberg, and Klein (2005) used a neural network simulation to show that attentional split in more than one location is more likely to be observed when the distance between the attended locations is relatively high compared to the spread of each individual attentional distribution. In other words, it seems that the distance between the attended locations is crucial for observing or not effects indicative of attentional split. Furthermore, we propose that another decisive factor for attention being split in multiple locations might be the presence or absence of frames, objects or, in general, physical stimuli at the to-be-attended locations. Specifically, consistently with many findings suggesting a parallelism between endogenous and exogenous mechanisms on one side and space-based and object-based on the other side (Lauwereyns, 1998; Macquistan, 1997), we hypothesize that it might be possible to split the focus only when attention is automatically captured by objects (exogenous attention), but not when it is voluntarily directed to empty locations (endogenous attention). 1.1. The present study According to the above-mentioned literature, in the present study we explored how the distance between endogenous and exogenous multiple cued locations affects the spatial distribution of attention, by studying its effects on a further level of information processing than perception. Specifically, following the logic suggested by Funes et al. (2007) to dissociate exogenous from endogenous attention, we analyzed their differential effects on VSWM information encoding. At this aim we used a task very similar to that of Botta et al. (2010), combining a cuing paradigm with a task involving identification in VSWM. With this new task the presentation of two endogenous (Experiment 1) or exogenous (Experiments 2 and 3) cues was followed by a memory array consisting of eight letters instead of eight colored squares. After a short interval, a location was probed for participants to report the identity of the letter that was presented there in the preceding memory array (see Fig. 1). The stimuli were circularly arranged in such a way that we could precisely control the distance between cued and probed locations. The main reason for this modification of Botta et al.'s paradigm, shifting from recognition to recall, was to increase task difficulty. In fact, according to Jans, Peters, and DeWeerd (2010) setting an appropriate task difficulty is a necessary condition to study divided attention. Nonetheless as this represents an important procedural change of Botta et al.'s study, we firstly replicated their main results in a preliminary study by using a single cue. A description of two experiments with this new procedure, one with a single endogenous and the other with a single exogenous cue is provided in Appendix A. Since the main results of our previous study were perfectly replicated, we used this new procedure in the current double-cue experiments.


3

Fixation (900 ms) 1500 ms delay Endogenous cue (300 ms)

+

Exogenous cue (100 ms) 50 ms delay

1-7

+

Memory array (100 ms) V

N

T

900 ms delay

V +

L

B R

Test array (10000 ms)

F

+

Fig. 1. Example of the procedure used in the experiments.

The use of two cues allowed several ways of examining the spatial distribution of the attentional bias on VSWM. On the one hand, as in Botta et al.'s study, we analyzed the effect of distance by considering the relative locations of the cues and the probe. Moreover, we could further examine this effect and test whether attention can be split, by just taking into account performance at the two attended locations and thus analyzing how it was affected by the distance between the two cues. We expected two different patterns of results depending on whether attention was either exogenously captured by peripheral cues or endogenously indicated by central symbolic cues. More precisely, coherently with the hypothesis that endogenous attentional orienting is based on a space-based coding system, in the case of endogenous cues we hypothesized a reliable effect of distance between attended locations, leading to better performance for contiguous than for noncontiguous attended locations. Contrariwise, we expected negligible effects of distance between attended locations following exogenous peripheral cues. In support of this hypothesis it is important to note that one of the main differences between endogenous and exogenous attentional paradigms is that while in the former central symbolic cues usually indicate locations, in the latter ones attended locations are in fact occupied and stimulated by objects (the peripheral cues) abruptly presented and automatically capturing our attentional resources. More importantly, in most of the paradigms that have been used to show divided attention, the to be attended locations do in fact coincide with two external singleton stimuli (i.e., peripheral cues; see Awh & Pashler, 2000; Bichot et al., 1999). It is worth reminding that, differently from endogenous attention, exogenous attentional orienting has been often conceived as operating in an object-centered frame of reference based on perceptual grouping laws (Boi, Vergeer, Ogmen, & Herzog, 2011). Concerning this last point, we also studied perceptual grouping effects by comparing performance at locations inside (i.e., in between) vs. outside the cued locations (see Fig. 2f). According to the event integration hypothesis of cueing effects (Lupiáñez, Mártín-Arévalo, & Chica, 2013; Lupiáñez, Ruz, Funes, & Milliken, 2007) we assume that the information carried out by the two peripheral cues will be grouped in the same object-file, thus performance being facilitated within the whole group, including the central locations of the group in which no

cue is presented. Specifically we predicted a better performance for targets presented inside the cues (i.e., in between them) than for targets presented outside the cues (i.e., not between them), but only when the two cued locations are sufficiently close (see Design and Data Analysis, Between-Cues Probe Position factor). In other words, when the cues are separated by an uncued location we expected performance in that location to be also facilitated as a consequence of an object-based grouping effect. Similar effects of perceptual organization in VSWM, which are perfectly consistent with the Gestalt principle of closure, have been already observed by Woodman, Vecera, and Luck (2003). However, we predicted these effects to occur only for exogenous attentional orienting. In contrast, for endogenous attention we expected no grouping effects, but rather performance being affected by the relation between the cued locations and the specific location of the probe. More exactly, we hypothesized that following endogenous cues, response accuracy would depend on the probe being presented in a different or in the same visual quadrant as one of the cues. We therefore predicted a significant effect of meridians not only when they were interposed between one of the cues and the probe, but also when they separated the two attended locations. Finally, we manipulated cue–target predictability orthogonally to cue type, in order to dissociate its effects from the effects produced by the way attention is oriented (endogenously by central/symbolic cues vs. exogenously by peripheral cues). 2. Experiment 1: double endogenous cues 2.1. Method 2.1.1. Participants Thirty-six psychology students from the University of Granada participated in the study. Thirteen of them participated in the unpredictive condition (4 males, mean age 21, ranging from 18 to 35 years), 10 in the 50% predictive condition (4 males, mean age 25, ranging from 21 to 37 years) and 13 in the 75% predictive condition (1 male, mean age 23.3, in the range of 17–53). They were naïve as to the purpose of the study, which lasted for approximately 40 min. All the experiments were conducted in accordance with the ethical

4


Cue-to-Cue Distance (CC-D) Cues-to-Probe Distance (CP-D)

a

CP-D1

e

CP-D2

Cued (CP-D0)

Between-Cues Meridian Crossing

CP-D1

Cued (CP-D0)

CP-D3

CP-D3 Horizontal Crossing

CC-D1 Cued (CP-D0)

b

CP-D2

CP-D3

CP-D1

CP-D2

Cued (CP-D0)

CP-D1

Cued (CP-D0)

CP-D3

CP-D1

f

CP-D1

Between-Cues Probe Position

CP-D2

Cued (CP-D0)

Outside different quadrant

CP-D1

Inside

CP-D1

CP-D3

Cued (CP-D0)

CP-D3

CC-D2

Cued (CP-D0)

CP-D1

CP-D2

Ouside same quadrant

CP-D2

!"

c

CP-D1

Cued (CP-D0)

g Between-Cues Probe Position

Cued (CP-D0) Inside different quadrant

CP-D1

CP-D1

Inside different quadrant

CP-D1 Outside same quadrant

CC-D3

Cued (CP-D0)

d

CP-D2

CP-D1

CP-D2

CP-D2

CP-D1

Cued (CP-D0)

CP-D3


CP-D2

h Cues-Probe Meridian Crossing

No Crossing CP-D1

Cued (CP-D0) Cued (CP-D0)

CP-D1

CC-D4 Cued (CP-D0)

CP-D1

No Crossing

CP-D1

CP-D1

Horizontal Crossing

CP-D1 CP-D1

CP-D2

Cued (CP-D0)

Vertical Crossing

Fig. 2. Schematic representation of the independent variables considered in the study. The first column represents the different levels of Cue-to-Cue Distance (CC-D), from the lowest (a) to the highest (d). Note that the Cues-to-Probe Distance (CP-D) changes depending on the Cue-to-Cue Distance. The second column represents a graphical example of the specific variable associated to each condition of Cue-to-Cue Distance: (e) Between-Cues Meridian Crossing (an example of horizontal crossing in the graph), associated to CC-D1; (f) and (g) Between-Cues Probe Position associated to CC-D2 and CC-D3 respectively; (h) Cues–Probe Meridian Crossing associated to CP-D1 for all levels of Cue-to-Cue Distance.


guidelines laid down by the Department of Experimental Psychology, University of Granada, in accordance with the ethical standards of the 1964 Declaration of Helsinki.

2.1.2. Stimuli The stimuli were displayed on a light gray background on a 17 in. LCD video monitor (refresh rate = 60 Hz) located in a dark and quiet room. The distance between the participant's head and the video monitor was approximately 60 cm. A black fixation was continuously displayed at the center of the screen. As illustrated in Fig. 1, the stimuli were presented at eight locations that were evenly spaced around an imaginary circle with a radius of approximately 7.5° that was centered at fixation. Each memory array consisted of a .95° × 1.43° letter at each of the eight possible equidistant locations, with two letters at each quadrant. Letters were upper case, randomly picked from the set BCDFGHJKLNPQRSTVZ (i.e., all consonants in Spanish alphabet apart from M, Ñ and W, which were not used in order to avoid too much confusion between letters). As endogenous cues, given that each of the eight locations coincides approximately with eight of the hours located on the analogical clock, we used 2 of 8 digits associated to them: 1, 2, 4, 5, 7, 8, 10, and 11 (see Fig. 1). The test arrays consisted of one probe (see Fig. 1) which was a square outline drawn in black, subtending a visual angle of 1.62° × 1.62°, located in one of the eight possible positions. In the unpredictive condition, the cues were totally unpredictive about the probe position (i.e., the probe could be presented equiprobably on each possible position, independently of the position of the cues). In the unpredictive, 50% predictive and 75% predictive conditions the two exogenous cues correctly indicated the probe position respectively in 25% of the trials (i.e., each cue was 12.5% valid), 50% of the trials (i.e., each cue was 25% valid) and in 75% of the trials (i.e., each was 37.5% valid).2

2.1.3. Procedure Each trial began with the presentation a central fixation cross. After 1500 ms two central numbers indicating two of eight possible locations appeared for 300 ms. After a 50 ms blank period the memory array was presented for 100 ms. This was followed by a 900 ms blank period and then by the test array in which a single probe appeared for 10,000 ms or until the participant's response. Participants were required to report the identity of the letter presented at the probe location in the preceding memory array. The cue-to-cue distance (CC-D) could vary from 1 (when the cues indicated two contiguous locations) to 4 (when the cues indicated opposite locations) (see Fig. 2a–d). The cues-to-probe distance (CP-D) changed depending on CC-D (0, 1, 2 and 3 for CC-D1 and CC-D2 and 0, 1 and 2 for CC-D3 and CC-D4) (see Fig. 2a–d). Eight blocks of 56 and 84 trials were presented for a total of 448 and 672 trials respectively in the unpredictive condition3 and in the two predictive conditions (with respectively 128 and 192 trials for CC-D1, CC-D2, CC-D3 and 64 and 96 for CC-D4).4

2 Note that the three conditions of Predictability (unpredictive, 50% predictive and 75% predictive) were in fact run as separated experiments, although they were completed in the same period (two months approximately). Since participants were not randomly assigned to one of the three conditions, they should technically be treated separately as different experiments. Nonetheless in order to simplify and facilitate the reading of present work we have considered Predictability as a between participants factor. 3 In this condition of predictability, even though cues were not predictive, participants were told that it was very important to attend to the cued locations. In this way we could dissociate the nature (symbolic) of the cue from its predictability. 4 Note that the total number of needed trials was calculated starting from a minimum number of trials for each invalid condition, thus producing an unbalance in the total number of trials for the three predictability conditions. Moreover, the use of 8 locations inevitably gives place to differences in the total number of trials of each level of cue-to-cue distance.

5

2.1.4. Design and data analyses Data analysis was divided in four sections, in all of them considering the percentage of correctly reported letters as dependent variable (see Table 1 for a resume of the analysis adopted to answer to each experimental question). As a first step we performed an ANOVA on the data derived from CP-D1 with the factor of Cues–Probe Meridian Crossing (no-crossing, horizontal and vertical crossing; see Fig. 2 h) to study the meridian effect between cues and probe. If the meridian crossing effect was not significant a further analysis was performed on the within factor of CP-D (0, 1, 2 and 3). If instead the meridian crossing effect was significant we divided the analysis of the cues-to-probe distance in two further analyses depending on the presence of a meridian: (a) an ANOVA for the conditions in which no meridian was crossed, with the within participants factor of CP-D (CP-D0 vs. CP-D1 No Crossing) and the between factor of Predictability (unpredictive, 50% predictive and 75% predictive); (b) an ANOVA for all conditions of meridian crossing, with the within-participants variable of CP-D (1, 2 and 3) and the between factor of Predictability (unpredictive, 50% predictive and 75% predictive). In the case of significant effect of CP-D, a trend analysis was carried out. To assess whether the cuing effect changes with the cue-to-cue distance and whether the attentional bias for storage into VSWM is affected by splitting the focus in two noncontiguous spatial locations, as a subsequent step we studied the effect exerted by distance between the two cued locations. At this aim we considered just response accuracy data derived from the cued-location trials (CP-D0; i.e., trials in which the probe is presented at the location of one of the cues). Also here as a first step, possible effects in correspondence with meridians were checked. Specifically, an analysis was performed on the data from CC-D1 with the within-participants factor of Between-Cues Meridian Crossing (no-crossing, horizontal and vertical; see Fig. 2e) and the between participants factor of Predictability. Depending on the results of this first analysis, the data derived from CC-D1 no-crossing condition were excluded or not from the following analysis of the within participants factor of CC-D (1, 2, 3 and 4) and the between participants factor of Predictability (unpredictive, 50% predictive and 75% predictive). In the case of significant effect of CC-D, trend analysis was carried out. Finally, we carried out further analyses to differentiate between intervening locations (i.e., in between the cues) and external locations (i.e., outside the cues). These analyses were performed with the data derived from CC-D2 and CC-D3 which constituted the two levels of cue-to-cue distance at which it is possible to differentiate between inside and outside cue locations. Moreover in the case of a significant main effect of Cues–Probe Meridian Crossing, we considered whether the specific external or intervening location was presented in the same or in a different quadrant than the cued locations. More specifically in the case of CC-D2 we performed an ANOVA on the data derived from CP-D1 on the within participants factor of Between-Cues Probe Position (inside-cues/same quadrant, outside-cues/same quadrant, outside-cues/different quadrant) and the between participants factor of Predictability. In the case of CC-D3 we performed a three-way ANOVA with the within participants factors of Between-Cues Probe Position (inside-cues vs. outside cues) Quadrant (same vs. different) and the between participants factor of Predictability5 (see Fig. 2f–g). Given the small differences between the sample sizes of the three predictability conditions, the default Type-III sum of squares was used to prevent confounding due to unequal group-sizes. 2.2. Results 2.2.1. Cues-to-Probe Distance (CP-D) The first analysis showed a significant effect of Cues–Probe Meridian Crossing, F(2, 66) = 16.2, p b .0001, η2p (partial eta-squared) = .33 5 Note that for CC-D2 the analysis is different as in this case inside cue condition is always within the same quadrant of one of the cues.

6


Table 1 The present table summarizes the main experimental questions treated in our study (first column), the variables involved to answer these questions (second column), and the figures representing the variables (third column). Questions

Variables

Figures

Do meridians affect the distribution of attentional bas in VSWM?

Cues–Probe Meridian Crossing (No-Crossing, Horizontal Crossing, Vertical Crossing) Between-Cues Meridian Crossing (No-Crossing, Horizontal Crossing, Vertical Crossing) Cues-to-Probe Distance (CP-D0, CP-Dl, CP-D2 and CP-D3)

Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig.

Do meridians affect the allocation of attention in two different locations? What is the effect exerted by the cue–probe distance on the distribution of the attentional bias? Does the distance between cued locations affect the attentional distribution? Are there costs in splitting the attentional focus? Is the attentional bias affected by perceptual grouping? Are inside cue locations better than outside cue locations? What is the effect of the cue–target predictability?

Cue-to-Cue Distance (CC-D1, CC-D2, CC-D3 and CC-D4) Between-Cues Probe Position (inside, outside) Predictability (unpredictive, 50% predictive and 75% predictive)

(see Fig. 3a). Bonferroni corrected pairwise comparisons indicated a significant better performance on no-crossing than on horizontal and vertical crossing conditions, p = .001 and p b .0001, which did not differ from each other, p = .13. The main effect of Predictability as well as its interaction with Cues–Probe Meridian Crossing, failed to reach significance, respectively p = .79 and p = .27. Concerning the distance effect in the absence of any meridian, participants were significantly better at CP-D0 than at CP-D1 No Crossing, F(1, 33) = 75.7, p b .0001, η2p = .69. Also in this case the main effect of Predictability as well as the interaction were not significant, respectively p = .9 and p = .075. Furthermore for the conditions of CP-D (1, 2 and 3) in which a meridian crossing was present in all cases, the analysis revealed a significant effect of distance, F(2, 66) = 35.1, p b .0001, η2p = .51. Trend analysis showed a significant linear component, p b .0001, indicating that performance gradually decreased with the distance (see Fig. 3a). No other significant main effects or interactions were observed, all ps N= .4. 2.2.2. Cue-to-Cue Distance (CC-D) Coherently with the above results related to the Cues–Probe Meridian Effect, we observed a significant main effect of BetweenCues Meridian Crossing, F(2, 66) = 19.3, p b .0001, η2p = .36. Response accuracy resulted better on no-crossing than on horizontal and vertical crossing conditions, both ps b .0001, which in turn were not significantly different from each other, p = .99 (see Fig. 3b). The main effect of Predictability as well as the interaction failed to reach significance, both Fs b 1. The subsequent analysis on the data derived from the cued trials (CP-D0) showed a significant effect of CC-D (1,6 2, 3 and 4), F(3, 99) = 12.4, p b .0001, η2p = .27. The significant linear component of the effect, p b .001, indicated that performance gradually decreased with the distance until CC-D2; then it reached an asymptote as indicated by a significant quadratic component, p = .005, and by the absence of significant differences between CC-D2 and CC-D4, p = .34. No other effects or interactions were observed, both F b 1. 2.2.3. CC-D2: Between Cues Probe Position The analysis revealed a significant effect of Between Cues Probe Position (inside-cues, outside-cues/same quadrant, outside-cues/different quadrant), F(2, 66) = 26.9, p b .0001, η2p = .44 (see Fig. 3c). Bonferroni corrected pairwise comparisons showed that when the probe was located outside the cues and in a different quadrant, performance was significantly worse than on both outside same quadrant trials and inside cue trials, both p b .0001. Importantly there were no differences between outside same quadrant trials and inside cues ones, p = .13. The main effect of Predictability and its interaction with Between Cues Probe Position failed to reach significance, respectively p = .76 and p = .22. 6 Note that as the Cue–Probe Meridian Effect was significant, we excluded no-crossing trials from CC-D1.

2h (graphic explanation) 3a (results) 2e (graphic explanation) 3b (results) 2a–d (graphic explanation) 3a (results) 2a–d (graphic explanation) 3b (results) 2f–g (graphic explanation) 3c (results) 4 (results)

2.2.4. CC-D3: Between Cues Probe Position The 2 × 2 × 3 analysis with the factor Between Cues Probe Position (inside vs. outside), Quadrant (same vs different) and Predictability, showed a significant main effect of Quadrant, F(1, 33) = 16.8, p = .0002, η2p = .33 (see Fig. 3c). Nonetheless a post-hoc analysis (Bonferroni corrected pairwise comparisons) of the significant interaction between Quadrant and Predictability, F(2, 33) = 4.7, p = .015, indicated that same quadrant trials were significantly better than outside trials only in the unpredictive condition, p = .0001. The trend was similar in the other two conditions of predictability (50% and 75%) but failed to reach significance, respectively p = .1 and p = .7. No other significant effects or interactions were observed, all Fs b 1.

2.3. Discussion Overall, the results of Experiment 1 can be synthesized as follows: a) less accurate responses in meridian crossing than in no crossing trials both when the cue and the probe and when the two cues were in different quadrants as compared to when they lay on the same quadrant, b) a gradual decrease of accuracy depending on the Cues-to-Probe distance with the minimum at CP-D3 (see Fig. 3a), c) better performance when the two cues were contiguous (CC-D1) than when they were not (CC-D2, CC-D3, and CC-D4) (see Fig. 3b), and d) in general, better performance for probes presented in the same quadrant than when they appeared in a different quadrant as one of the cues, but independently on the probe being presented in between the two cues or outside them (see Fig. 3c). Taken together, the presence of a reliable effect of meridians as well as the reliable cost of attentional split, suggest, coherently with our theoretical proposal, that spatial distribution of endogenous attention is likely to be characterized by a unipolar spatial system of reference based on location. Moreover the absence of significant modulations by the cue predictability indicates that such a distribution is more related to the symbolic nature of the cue than to its validity. This last conclusion will be further supported by the results of the next experiment, in which peripheral cues were used instead, while also manipulating cue predictability.

3. Experiment 2: double exogenous cues 3.1. Method 3.1.1. Participants A new group of thirty-six psychology students from the University of Granada participated in the study. Twelve participated in the unpredictive condition (2 males, mean age 20.9, in the range of 21–26), 11 in the 50% predictive condition (1 male, mean age 22.5, in the range of 18–32) and 13 in the 75% predictive condition.


a

100

Endogenous(Exp.1) No Crossing

90 80

b

Exogenous(Exp.2)

No Crossing

90 80

Exogenous(Exp.2)

Meridian Crossing (vertical or horizontal)

70

Correct (%)

70

Correct (%)

Endogenous(Exp.1)

100

Meridian Crossing (vertical or horizontal)

7

60 50 40

60 50 40

30

30

20

20

10

10

0

0 CP-D0

CP-D1 NC

CP-D1

CP-D2

CP-D3

CC-D1 NC

CC-D1

Cues-to-Probe Distance Cues-Probe Meridian Crossing

CP-D2

Between-Cues Meridian Crossing

No Crossing CP-D1

CP-D2

CC-D2

CC-D3

CC-D4

Cue-to-Cue Distance

Cued (CP-D0)

CP-D1

CP-D2

Cued (CP-D0)

CP-D3 Horizontal Crossing

CP-D1

No Crossing

Horizontal Crossing

CP-D1

Cued (CP-D0)

CP-D1 Vertical Crossing

Cued (CP-D0)

CP-D2

CP-D1

Endogenous(Exp.1)

c

CP-D3

Exogenous(Exp. 2)

100 CC-D2

90

CC-D3

80

Correct (%)

70 60 50 40 30 20 10 0 Inside


Outside/SQ

Outside/DQ

Inside/SQ

Inside/DQ

Outside/DQ


Cued (CP-D0) Outside different quadrant

Cued (CP-D0) Inside different quadrant

Inside

CP-D1

Inside different quadrant

Cued (CP-D0)

CP-D3

Cued (CP-D0)


Outside/SQ


CP-D2

CP-D1 Outside same quadrant CP-D3 CP-D2

Fig. 3. Results of Experiment 1 (endogenous attention) and Experiment 2 (exogenous attention): a) Mean percentage of accuracy response as a function of Cues-to-Probe distance (CP-D) and Cues–Probe Meridian Crossing; b) Mean percentage of accuracy response for the cued trials (CP-D0) as a function of the Cue-to-cue distance (CC-D) and Between-Cues Meridian Crossing; c) Mean percentage of accuracy response as a function of Between-Cues Probe Position at CC-D2 and CC-D3.

3.1.2. Stimuli and procedure The stimuli and procedure were identical to those used in Experiment 1, with the exception that now two exogenous cues consisting of two black circles were presented for 100 ms in two of the eight possible locations (see Fig. 1). 3.1.3. Design and data analyses Data analysis was as in Experiment 1.

3.2. Results 3.2.1. Cues-to-Probe Distance (CP-D) The first analysis on Cues–Probe Meridian Crossing showed that this factor, as can be observed in Fig. 3a, did not significantly affect response accuracy, F b 1, not even its interaction with Predictability, p = .2. We observed instead a significant effect of Cues-to-Probe Distance (0, 1, 2 and 3) F(3, 99) = 169, p b .0001, η2p = .83. The significant linear

8


component of the effect, p b .0001, indicated that performance gradually decreased with the distance until CP-D2 where it reached an asymptote (see Fig. 3a) as attested by a significant quadratic component, p b .0001, and by the absence of significant differences between CP-D2 and CP-D3. More interestingly, the analysis revealed a significant interaction between Predictability and Cues-to-Probe Distance, F(6, 99) = 14.5, p b .0001, η 2p = .46. Particularly, pairwise comparisons indicated on the one side that response accuracy at the cued location was better when the cue was predictive (75% and 50%) than when it was unpredictive p b .0001; instead, at CP-D1 performance was better when the cue was unpredictive than when it was predictive, p = .002. In other words the cueing effect was larger with predictable cues than with unpredictable cues, p b .001, due to a larger benefit at the cued location and a larger cost at the uncued locations (see Fig. 4).

3.2.2. Cue-to-Cue Distance (CC-D) Again the main effect of Between-Cues Meridian Crossing as well as its interaction with Predictability, failed to reach significance, respectively p = .1 and p = .55. The subsequent analysis on the data derived from the cued trials showed a significant effect of Cue-to-Cue distance, F(3, 99) = 3.5, p = .01, η2p = .097. The polynomial trend analysis revealed a significant quadratic component, p = .04, with a minimum at CC-D2. There were no significant differences between CC-D1, CC-D3 and CC-D4, all ps N .4 (see Fig. 3b). We also observed a significant effect of Predictability, F(2, 33) = 13.2, p b .0001, η2p = .44, showing the pattern just described. Accuracy was lower in the unpredictive condition than in both the 50% and the 75% predictive conditions, respectively p = .0007 and p b .0001, which in turn did not differ from each other, p = .28. The interaction between CC-D and Predictability failed to reach significance, F b 1.

3.2.3. CC-D2: Between Cues Probe Position The ANOVA revealed a significant main effect of Between Cues Probe Position (inside vs. outside), F(1, 33) = 19.6, p = .0001, η2p = .37, indicating that performance at inside cue location was significantly better than at outside cue locations (see Fig. 3c). The interaction between Predictability and Between Cues Probe Position was not significant, F b 1.

3.2.4. CC-D3: Between Cues Probe Position The main effect of Between Cues Probe Position and its interaction with Predictability failed to reach significance, respectively, F(1, 33) = 2.3, p = .13 η2p = .06 and F b 1.

25% (unpred.)

50%

3.3. Discussion The results of the present experiment indicated that, in contrast to Experiment 1, when attention is captured by peripheral cues the attentional bias on VSWM seems to be slightly affected by the distance between cues. Performance at cued locations, instead of linearly decreasing with CC-D, as it would be expected by a “unitary” model of attention, was characterized by an overall flat curve with a point of minimum at CC-D2. There were no significant differences between cued location trials at CC-D1, CC-D3 and CC-D4 suggesting that the attentional bias effect in VSWM does not necessarily decrease depending on the spatial contiguity of the cued locations. When the cues were separated but close enough (CC-D2), performance was significantly better at inside than outside cue locations. This pattern of results is particularly crucial to understand the point of minimum observed in cued trials at CC-D2. In fact it is possible that the decrease of performance at CC-D2 is the consequence of an increase of the size of the attentional “focus”, due to grouping processes encompassing the ensemble of the two cued locations and the inside cue location (see General discussion). The data analysis also revealed a strong influence of Predictability, which was absent in Experiment 1 with central cues (see Fig. 4). For cued location trials performance was significantly better in the predictive conditions than in the unpredictive condition. The opposite pattern was observed for uncued trials in which response accuracy was always better in the unpredictive than in both predictive conditions. In other words, predictability increased the attentional facilitation (Cued–Uncued difference) over VSWM encoding. If we consider the predictability-dependent accuracy increase at the cued locations as attentional benefits and, on the other hand, the predictability-dependent accuracy decrease at the uncued locations (CP-D1, CP-D2 and CP-D3) as costs, our data indicate that predictability augments both costs and benefits coherently with Doricchi, Macci, and Macaluso (2010). This issue will be discussed further in the General discussion.

4. Experiment 3 The results of Experiment 2 suggested that performance at cued locations, instead of linearly decreasing with CC-D, was characterized by an overall flat curve with a point of minimum at CC-D2. We interpreted this effect as due to a grouping process of the three letters enclosed between the two cued locations. This logically leads to an increase of the size of the attentional focus which might have as a consequence the drop in performance at the two cued locations specifically observed at this distance (the point of minimum at CC-D2). On the

75%

25% (unpred.)

100

75%

100

Endogenous (Exp.1)

90

Exogenous (Exp.2)

90

80

80

70

70

Correct (%)

Correct (%)

50%

60 50 40 30

60 50 40 30

20

20

10

10

0

0 CP-D0

CP-D1 NC

CP-D1

CP-D2

CP-D3

CP-D0

CP-D1 NC

CP-D1

CP-D2

Fig. 4. Mean percentage of accuracy as a function of Cues-to-Probe Distance and Predictability in Experiment 1 and Experiment 2.

CP-D3


other hand, the absence of significant differences between cued trials at CC-D1, CC-D3 and CC-D4 suggested that the spatial distribution of the attentional bias is not necessarily affected by the separation of the cues in two noncontiguous regions. Nonetheless since we did not monitor ocular movements we cannot totally exclude the possibility that participants used eye movement strategies during the experiment. In the present experiment we handled this problem by using an eyetracker to monitor eye movements. Moreover we reduced the exposure of the memory array as well as the SOA between the cue presentation and the memory array so to avoid possible attentional shifts from one cue to the other, which might explain the observed pattern of data in Experiment 2. 4.1. Method 4.1.1. Participants Fifteen psychology students from the University of Granada participated in the study (mean age 23, in the range of 18–50). They were naïve as to the purpose of the study, which lasted for approximately 40 min.

9

4.2.2. Cue-to-Cue Distance (CC-D) Coherently with Experiment 2a the first analysis showed that the main effect of Between-Cues Meridian Crossing was far from significance, F b 1. The analysis of the distance between cues converged with the analogous analysis in Experiment 2. The CC-D effect was statistically significant, F(3, 42) = 5.7, p = .002, η2p = .29. A significant quadratic component, p = .01, confirmed a point of minimum at CC-D2. There were no significant differences between CC-D1, CC-D3 and CC-D4, all ps N .12 (see Fig. 5B). 4.2.3. CC-D2: Between Cues Probe Position Similarly as in Experiment 2 performance was better at inside cue than at outside cue locations F(1, 14) = 5.1, p = .03, η2p = .26 (please see Fig. 5C). 4.2.4. CC-D3 As expected, at CC-D3 there were no significant differences between inside and outside cue locations, F(1, 14) = 1.2, p = .27, η2p = .08. 4.3. Discussion

4.1.2. Apparatus Eye movements were recorded using a SMI iViewRED 250, tracking binocularly at 250 Hz with a spatial resolution of 0.1°, and a maximum average gaze position error of 0.5°. 4.1.3. Stimuli and procedure The stimuli and procedure were identical to those used in Experiment 2, unpredictive condition, except for the following. The temporal presentation of the cues as well as the delay between the cues and the memory array was reduced respectively to 50 and 16 ms, thus leading to a SOA of 66 ms. Moreover the memory array exposure was reduced from 100 to 50 ms. 4.1.4. Design To control for eventual confounding due to the different total number of trials between the unpredictive condition and the predictive conditions of Experiment 2, the total number of trials in this new experiment was made equal to that of Experiment 2's predictive conditions. Specifically, eight blocks of 84 trials were presented for a total of 672 trials respectively (with respectively 192 trials for CC-D1, CC-D2, CCD3 and 96 for CC-D4). 4.2. Results Trials in which gaze deviated more than 2° from fixation were discarded from the analysis (5.4%). 4.2.1. Cues-to-Probe Distance (CP-D) The first analysis on Cues–Probe Meridian Crossing revealed that, coherently with Experiment 2, there were no differences between Nocrossing condition and both horizontal and vertical crossing conditions, both ps N .25 (see Fig. 5A).7 As in Experiment 2, the analysis revealed a significant effect of Cuesto-Probe Distance (0, 1, 2 and 3), F(3, 42) = 56.8, p b .0001, η2p = .8. Once again a significant linear component of the effect, p b .0001, indicated that performance gradually decreased until CP-D2; then it reached an asymptote as attested by a significant quadratic component, p b .0001, and by the absence of significant differences between CP-D2 and CP-D3, p N .3. 7 There was a significant main effect of Cues–Probe Meridian Crossing, F(2, 28) = 3.68, p = .038, η2p = .20. Post-hoc comparisons (Bonferroni test) revealed that horizontal crossing condition was significantly better than vertical crossing condition, p = .03. It is not clear why this difference was observed, but in any case it is important to note that there were no differences between meridian crossing and no crossing conditions.

Overall the results of the present experiment confirmed those observed in Experiment 2, thus excluding explanations of the main pattern of results in terms of ocular and/or attentional movement strategies. 5. General discussion The primary goal of this study was to thoroughly investigate the differences between exogenous and endogenous mechanisms in the spatial distribution of attentional bias on VSWM when two locations are cued. At this aim, we used a paradigm in which two spatial locations were either exogenously (by peripheral cues), or endogenously (by means of central/symbolic cues) cued, to study the effects of exogenous vs. endogenous attention on an identification task in VSWM. In agreement with theoretical models considering endogenous attention as a space-based mechanism, when attention was simultaneously directed to two different locations by two endogenous central cues, we expected a unipolar distribution of attentional effects, characterized by meridian crossing effects and reliable costs of splitting the attentional focus in two halves. Oppositely, for exogenous cues, in accordance with object-based theoretical views of exogenous attentional mechanisms (Botta, Lupianez, & Sanabria, 2013), we predicted a spatial distribution of attentional bias characterized by negligible effects of distance between attended locations (i.e., no costs for attentional split in two noncontiguous locations), no effects of spatial anisotropy and significant grouping effects at short distances between the two attended locations. Overall our results undoubtedly supported our hypotheses, by showing a clear difference between Experiment 1 (with central cues) and Experiments 2 and 3 (with peripheral cues). This differential pattern of results for central vs. peripheral cues is well in agreement with many previous studies claiming a dissociation between endogenous and exogenous attentional processes (Funes et al., 2007; Klein & Shore, 2000). One of the foremost observed distinction concerns the role of the vertical and horizontal meridians in the spatial distribution of the attentional effects. In line with our previous study (Botta et al., 2010) exclusively endogenous attention was affected by the vertical and horizontal axes of the visual field, as indicated by the presence of significant differences between meridian crossing and no-crossing conditions in Experiment 1. What is more, the present results suggest that the meridians do not simply represent a further cost in reorienting attention between cued and uncued positions (Rizzolatti et al., 1987), but they might constitute a “hindrance” in splitting the attentional focus in two different quadrants, as indicated by the systematic Between-Cues Meridian Crossing effect. This interpretation is further

10


a

b

100

90

90

No Crossing

80

Meridian Crossing (horizontal or vertical)

80

Meridian Crossing (horizontal or vertical)

No Crossing

70

Correct (%)

70

Correct (%)

100

60 50 40

60 50 40

30

30

20

20

10

10 0

0 CP-D0

CP-D1 NC

c

CP-D1

CC-D1 NC CC-D1

CP-D3

CC-D2

CC-D3

CC-D4

100 90

CC-D2

80

Correct (%)

CP-D2

CC-D3

70 60 50 40 30 20 10 0 Inside

Outside/SQ

Outside/DQ

Inside/SQ

Inside/DQ

Outside/SQ

Outside/DQ

Fig. 5. Results of Experiment 3: a) Mean percentage of accuracy response as a function of Cues-to-Probe distance (CP-D) and Cues–Probe Meridian Crossing; b) Mean percentage of accuracy response for the cued trials (CP-D0) as a function of the Cue-to-cue distance (CC-D) and Between-Cues Meridian Crossing; c) Mean percentage of accuracy response as a function of Between-Cues Probe Position at CC-D2 and CC-D3.

supported by performance being always better for probes appearing in the same quadrant as one of the two endogenously cued locations (inside cues and outside cues/same quadrant trials) than for probes appearing in a different quadrant. On the other side, the attentional bias exerted by endogenous cues seems to linearly decrease with distance, coherently with a simple attentional gradient model (Hughes & Zimba, 1985). Finally, results showed a reliable cost of attentional split, suggesting that endogenous attention cannot be divided simultaneously in two noncontiguous empty locations. A quite different pattern of results emerged in Experiment 2, where peripheral cues were used to simultaneously and exogenously capture attention at two different locations. Consistently with single cue studies, the results confirmed an effect of cues–probe distance characterized by a decrease of performance with the distance between the cues and the probe up to CP-D2 where it reached an asymptote. More importantly, and contrarily to endogenous cues, the cues– probe distance effect was significantly affected by predictability. Specifically attention facilitation (i.e. the difference between cued and uncued locations) reliably increased with predictability. As described in the discussion of Experiment 2, and shown in Fig. 4, the larger difference between cued and uncued location trials for the predictive conditions might be due to larger attentional benefits at the cued location, and larger costs at the uncued locations. Doricchi et al. (2010) (see also Lasaponara, Chica, Lecce, Lupianez, & Doricchi, 2011), showed that costs (but not benefits) increase with cue–target predictability, when automatic orienting cues were used (central arrows in this case). As peripheral cues also orient attention automatically, the costs they produce might also be increased by cue–target predictability. Future research

should further investigate whether attentional costs observed with central symbolic cues also increase with predictability. Regarding attentional benefits, it is also interesting that the effects of central cues were not affected by predictability, which might be counterintuitive. Nevertheless, what is counterintuitive is that cueing effects were observed for non-predictive central cues. It must be noted that participants were not instructed about predictability, but were instead clearly encouraged to orient their attention to the cued locations. In other words the effects of central cues depend on the intention to use the cue information and orient attention according to it. Once attention is endogenously oriented, predictability has no further effect. In contrast, in the case of peripheral cues participants were not told anything about the cue–target relationship. Attention was automatically captured at the cued locations, perhaps also in this case, independently of predictability. However, later processes acting after attentional capture, such as the active maintenance of attention at the cued location, or the use of cue– target integration for further processing (Chica et al., 2008), might explain the larger attentional benefits observed for predictive peripheral cues. On the other side, the distance between cues slightly modulated the encoding in VSWM. The function describing performance at cued locations was in fact characterized by an overall flat curve with a point of minimum at distance 2 (CC-D2) between the two cued locations. The absence of significant differences between contiguous (CC-D1) and noncontiguous locations at higher distances than CC-D2 (i.e., CC-D3 and CCD-4) suggests that attention can be simultaneously captured by two noncontiguous stimuli. Experiment 3, by monitoring eye


movements and reducing the cue-memory array SOA, allowed us to exclude an interpretation of this pattern of data in terms of either ocular or attentional movement strategies. The minimum observed at CC-D2, finds an interesting interpretation in the comparison between inside (i.e., in between cues) and outside cue locations. In fact, throughout all conditions of predictability, we observed that when the distance between cues is short (CC-D2) performance at the inside cue location was always better than at outside cue locations. This finding seems to be coherent with the data of Klein, Christie, and Morris (2005) who asserted that “an orienting response is programmed to the center of gravity or center of area of the element making up the cue” (p. 299), i.e., to the net vector of the different orienting directions. Nonetheless, as Klein et al. measured IOR instead of facilitation, a direct comparison is not allowed. In any case, an alternative interpretation of these results is that they are the consequence of a grouping process leading the ensemble of the three letters enclosed between the two cues to be perceived as a single perceptual group. This alternative explanation is consistent with the cue–target integration or event integration theory (Lupiáñez, Milliken, Solano, Weaver & Tipper, 2001; Lupiáñez et al., 2007). According to this theory, two spatio-temporally contiguous events, as the cue and the target at short SOAs, are encoded within the same event representation. In agreement with the object-file theory originally advanced by Kahneman, Treisman, and Gibbs (1992), spatial and temporally contiguous events are stored in the same object file by means of an updating process which involves the information carried out by the prior (the cue) and the current event (the probe). This happens when the spatiotemporal contiguity between two events is sufficiently high. Specifically, in our paradigm, when the probe was presented inside the cues, it is highly likely that the information held in these three spatial locations would be stored in the same object file as a consequence of an object-based grouping effect (Funes et al., 2007). Similar arguments have been recently stated to explain the effects of auditory cues on storing visual information into VSWM (Botta et al., 2013). Once the separation between cues is large enough, two object representations would be created instead, leading to clear effects of attentional split, in contrast to the prediction of a single vector coding proposed by Klein et al. (2005). Interestingly, the general pattern of results observed here for peripheral exogenous cues (Experiments 2 & 3), is perfectly coherent with a model recently proposed by Feng and Spence (2012). These authors observed that since attention is a resource-limited process (Kahneman, 1973) the subdivision of attention among more than one location should necessarily lead to poorer performance at the individual locations (McMains & Somers, 2005). They proposed that when more locations are simultaneously cued, attention is divided, creating many unimodal attentional distributions (Standage et al., 2005) and giving place to a joint distribution of attention defined as the mixture (in the usual statistical sense, e.g., McLachlan & Peel, 2000) of the individual attentional distributions. The mixture model proposed by Feng and Spence (2012) predicts that if the distance between the cued locations is relatively large (i.e., our CC-D3 and CC-D4), the attentional surface predicted by the mixture model is consistent with a multiple-foci theoretical perspective. However when there is only one cue, or multiple closely-spaced cues (i.e., our CC-D1 and CC-D2), the attentional surface predicted by the mixture model will be coherent with a single focus model, which in our view could be the consequence of a perceptual grouping process. Overall perhaps the safer conclusion that can be extracted from our data is that depending on the number of the intervening locations the system chooses the best strategy to optimize performance (single vs. multiple foci). If the peripheral cues/objects capturing attention are close enough to be perceptually grouped, attention would be deployed to the perceptual group, thus producing a unimodal spatial distribution of attention; nonetheless, if they are distant enough as to be perceived as different objects, the attentional focus would be divided to the different locations, giving place to a multimodal attentional distribution.

11

There are, however, other explanations alternative to an interpretation in terms of multiple attentional foci. One of the most interesting hypotheses is that attention takes the most convenient shape depending on the specific configuration of the multiple cues. Coherently with this idea a recent study indicates that perceptual grouping allows attending noncontiguous locations while suppressing capture from intervening locations (Kerzel, Born & Schönhammer, 2012). According with this view, our results could be perfectly interpreted in terms of cue–target perceptual grouping processes. As a final remark, it has been observed that an equal distribution of attentional resources between two locations can be maintained only for a limited time delay (Dubois, Hamker, & VanRullen, 2009). This is very interesting considering that attentional split was only observed with exogenous attention (peripheral cues) in our experiments. Note that exogenous attention has been often conceived as a rapid but short-lasting attentional mechanism (Müller & Rabbitt, 1989; Nakayama & Mackeben, 1989), on the one side, and as object-based (Boi et al., 2011; Kristjansson, Mackeben, & Nakayama, 2001; Macquistan, 1997) on the other side. We speculate that attention cannot be voluntary and endogenously allocated in more than one location, unless these locations are occupied by objects, or stream of objects (Müller et al., 2003). In contrast, attention can be simultaneously captured by objects or object configurations for a relative short delay of time, allowing early attentional benefits in information processing. After this initial spread of attention it might follow a later selection of one of the attentional foci for voluntary attentional allocation leading to a deeper stimulus elaboration. It remains for future studies to understand the deepness of processing achievable when attention is divided in multiple foci and the role exerted by grouping laws and temporal factors involved in this process. Acknowledgments This research was supported by research projects PSI2011-22416, to J.L. and eraNET-NEURON BEYONDVIS, EUI2009-04082, to J.L. and F.B. We would like to thank S. Cacace for programming assistance. Appendix A. Pilot experiment: single cue This pilot experiment had three main aims: 1) to test the variant of the Botta et al. (2010) task that will be used in all experiments of this paper, 2) to replicate Botta et al. (2010) results with one single cue, and 3) to serve as a baseline for a comparison with the subsequent double cue experiments. The main difference between the present task and Botta et al.'s task is that while in the previous study the cuing paradigm was combined with a change detection task, here we have combined it with a paradigm involving identification in VSWM to increase task difficulty. Setting an appropriate task difficulty represents an important criterion to study divided attention (Jans et al., 2010). A.1. Method A.1.1. Participants The group of participants consisted of 16 volunteers, psychology students from the University of Granada. Half of them (2 male, mean age 24.8, ranging from 21 to 34 years) participated in Experiment 1A, while the other 8 (3 male, mean age 27.5, ranging from 20 to 38 years) participated in Experiment 1B. All participants had a normal or corrected-to-normal visual acuity, normal color discrimination, and no history of neurological problems. They were naïve as to the purpose of the study, which lasted for approximately 40 min. All the experiments were conducted in accordance with the ethical guidelines laid down by the Department of Experimental Psychology, University of Granada, in accordance with the ethical standards of the 1964 Declaration of Helsinki.

12


A.1.2. Stimuli The stimuli were displayed on a light gray background on a 17 in. LCD video monitor (refresh rate = 60 Hz) located in a dark and quiet room. The distance between the participant's head and the video monitor was approximately 60 cm. A black fixation was continuously displayed at the center of the screen. As illustrated in Fig. 1, the stimuli were presented at eight locations that were evenly spaced around an imaginary circle with a radius of approximately 7.5° that was centered at fixation. Each memory array consisted of a .95° × 1.43° letter at each of the eight possible equidistant locations, with two letters at each quadrant. Letters were upper case, randomly picked from the set BCDFGHJKLNPQRSTVZ (i.e., all consonants in Spanish alphabet apart from M, Ñ and W, which were not used in order to avoid too much confusion between letters). The exogenous cue (Experiment 1B) was a black circle, subtending a visual angle of .65° × .65°, which was presented at one of the 8 possible locations, but was totally unpredictive about the probe position. As endogenous cues (Experiment 1A), given that each of the eight locations coincides approximately with eight of the hours located on the analogical clock, we used 8 digits associated to them: 1, 2, 4, 5, 7, 8, 10, and 11. In 75% of the trials the cue correctly indicated the upcoming probe location (i.e., valid trials); the remaining trials were instead invalid, and were equally subdivided between all combinations of cue location × probe location. The test arrays consisted of one probe which was a square outline drawn in black, subtending a visual angle of 1.62° × 1.62°, located in one of the eight possible positions. A.1.3. Procedure Each trial began with the presentation a central fixation cross. After 1500 ms in the endogenous experiment (Experiment 1A) a central digit indicating one of the eight possible locations was presented for 300 ms; in the exogenous experiment (Experiment 1B) a circle cue appeared for 100 ms at one of the eight locations. After a 50 ms blank period the memory array was presented for 100 ms. This was followed by a 900 ms blank period and then by the test array that appeared for 10,000 ms or until the participant response. The cue-to-probe distance (CP-D), that is the distance between the cue position and the probe position, could vary between 0 (CP-D0 or cued position) and 4 (CPD4) that represents the condition in which they are located on opposite sides of the imaginary circle. The participants had to report the identity of the letter presented in the probe location by pressing the corresponding letter on the computer keyboard. In the endogenous orienting experiment (Experiment 1A) a total of 896 trials were presented (with 64 trials for CP-D1, CP-D2, CPD3; 672 trials for CP-D0, or cued position, and 32 for CP-D4). The experiment was subdivided in eight blocks of 112 trials. In Experiment 1B six blocks of 64 trials were presented (i.e., 8 cue locations × 8 target locations), for a total of 384 trials (with 96 trials for CP-D1, CP-D2, and CP-D3, and 48 trials for CP-D0, or cued position, and CP-D4). The participants were allowed to rest for a few minutes between blocks. Before the start of the experiment participants performed 30 training trials. A.1.4. Design and analyses Only percentages of correctly reported letters were considered for the analyses. As CP-D1 is the only invalid condition in which there were both meridian crossing and no-crossing conditions, and equating distance is obviously necessary to study the meridian effect, we firstly performed a univariate ANOVA on the data derived from the CP-D1 condition, with the within-participants factor of Meridian Crossing (no-crossing, horizontal and vertical crossing). When the meridian crossing effect was not significant a further analysis was performed to investigate the distance effect. Another univariate ANOVA was conducted with the withinparticipants factor of Cue-to-Probe Distance (CP-D0, CP-D1, CP-D2, CPD3 and CP-D4) without considering meridian crossings.

When the meridian crossing effect was significant two further analyses were performed. On the one hand, we compared CP-D0 and CP-D1 (No-Crossing) conditions by a t-test to verify the effectiveness of the cue in the absence of meridian crossings. An ANOVA was performed with the within-participants factors of Meridian Crossing (vertical vs. horizontal) and Cue-to-Probe Distance (CP-D1, CP-D2 and CP-D3) to examine the interaction between them (we excluded the trials at CP-D4 and some of the trials at CP-D3 because they involve the crossing of both meridians). A.2. Results and discussion A.2.1. Experiment 1A (endogenous cue) The analysis of Meridian Crossing showed a highly significant effect, F(2, 14) = 12.39, p b .001. Planned comparisons pointed out a better accuracy in the no-crossing condition (31.6% correct) than in the horizontal (10.9%) and vertical (10.1%) crossing conditions, p b .002, which instead resulted statistically identical, p = .875. The t-test between D-0 and D-1 NC showed a significant better performance at CP-D0 (62.5% correct) than at CP-D1 NC, p b .004. The Cue-to-Probe Distance (1, 2, and 3) × Meridian crossing (horizontal and vertical) ANOVA revealed no interaction between the two factors, F b 1. The main effect of Meridian Crossing was also not significant F b 1. The main factor of Distance revealed instead a significant effect, F(2, 14) = 5.27, p b .02. Planned comparisons indicated a better accuracy at D-1 (10.5% correct) than at D-2 (4.2% correct), p b .03, which in turn did not differ from D-3 (4.2% correct). 8 A.2.2. Experiment 1B (exogenous cue) In contrast to Experiment 1A with endogenous cue, the analysis of Meridian Crossing with exogenous cues revealed no differences between no-crossing, horizontal and vertical meridian crossing conditions, F b 1. The analysis of Cue-to-Probe Distance, however, revealed a highly significant effect, F(4, 28) = 23.08, p b .0001. Post-hoc analysis revealed that performance decreased linearly with distance (CP-D0 65.6% correct, CP-D1 39% correct and CP-D2 24.4% correct), p b .005, as indicated by a significant linear component of the effect, p b .0009, and it reached an asymptote at CP-D2 as confirmed by a significant quadratic component of the effect, p b .007, and by the absence of significant differences between CP-D2 and CP-D3 (23.9% correct) and between CP-D3 and CP-D4 (23.9% correct). To sum up, only a distance effect (up to distance 2) was observed with exogenous cues, without any sign of meridian crossing effect. However, with endogenous cue, the distance effect coexisted with a much larger meridian crossing effect. Thus, we replicated in this new task the dissociation between exogenous attention and endogenous attention as regards their modulation over VSWM previously reported by Botta et al. (2010). Another important result observed in this experiment was that performance on locations distant from the cue was much lower with the endogenous (around 4%) than with the exogenous cue (greater than 20%). It is not clear whether this difference is due to the cue type or to its predictability. We will consider this issue in the main study. References Awh, E., & Jonides, J. (2001). Overlapping mechanism of attention and spatial working memory. Trends in Cognitive Sciences, 5, 119–126. Awh, E., & Pashler, H. (2000). Evidence for split attentional foci. Journal of Experimental Psychology: Human Perception and Performance, 26, 834–846. Awh, E., Vogel, E., & Oh, S. H. (2006). Interactions between attention and working memory. Neuroscience, 139, 201–208. 8 A further analysis with the factor Number of Crossings (D-3/1 meridian crossing, D-3/ 2 meridian crossing and D-4/2 meridian crossing) resulted not significant. The absence of difference between D-3/1 meridian crossing and D-3/2 meridian crossing revealed that meridians did not have an additive effect; moreover the performance at D-3/2 meridian crossing was not different to that at D-4 indicating no further distance effects.

F. Botta, J. Lupiáñez / Acta Psychologica 150 (2014) 1–13 Baldauf, D., & Deubel, H. (2008). Properties of attentional selection during the preparation of sequential saccades. Experimental Brain Research, 184, 411–425. Bichot, N., Cave, K., & Pashler, H. (1999). Visual selection mediated by location: Feature based selection of noncontiguous locations. Perception & Psychophysics, 61, 403–423. Boi, M., Vergeer, M., Ogmen, H., & Herzog, M. H. (2011). Non retinotopic exogenous attention. Current Biology, 21, 1732–1737. Botta, F., Lupianez, J., & Sanabria, D. (2013). Visual unimodal grouping mediates auditory attentional bias in visuo-spatial working memory. Acta Psychologica, 144, 104–111. Botta, F., Santangelo, V., Raffone, A., Lupianez, J., & Olivetti Belardinelli, M. (2010). Exogenous and endogenous spatial attention effects on visuo-spatial working memory. Quarterly Journal of Experimental Psychology, 27, 1–13. Bundesen, C. (1990). A theory of visual attention. Psychological Review, 97, 523–547. Carlson, T., VanRullen, R., Hogendoorn, H., Verstraten, F., & Cavanagh, P. (2007). Distinguishing models of multifocal attention: It's a matter of time. Journal of Vision, 7(9), 641. Carrasco, M., & McElree, B. (2001). Covert attention accelerates the rate of visual information processing. Proceedings of the National Academy of Sciences, 98, 5363–5367. Cave, K., & Bichot, N. (1999). Visuospatial attention: Beyond a spotlight model. Psychonomic Bulletin & Review, 6, 204–223. Chica, A. B., Botta, F., Lupiáñez, J., & Bartolomeo, P. (2012). Spatial attention and conscious perception interactions and dissociations between and within endogenous and exogenous processes. Neuropsychologia, 50, 621–629. Chica, A. B., Charras, P., & Lupiáñez, J. (2008). Endogenous attention and Illusory Line Motion depend on task set. Vision Research, 48(21), 2251–2259. Chica, A. B., Lasaponara, S., Chanes, L., Valero-Cabré, A., Doricchi, F., Lupiáñez, J., et al. (2011). Spatial attention and conscious perception: The role of endogenous and exogenous orienting. Attention, Perception, & Psychophysics, 73, 1065–1081. Doricchi, F., Macci, Silvetti, & Macaluso, E. (2010). Neural correlates of the spatial and expectancy components of endogenous and stimulus-driven orienting of attention in the Posner paradigm. Cerebral Cortex, 20, 1574–1585. Dubois, J., Hamker, F., & VanRullen, R. (2009). Attentional selection of non-contiguous locations: The spotlight is only transiently split. Journal of Vision, 9(5), 1–11. Eriksen, C. W., & St. James, J. D. (1986). Visual attention within and around the field of focal attention: A zoom lens model. Perception & Psychophysics, 40, 225–240. Feng, J., & Spence, I. (2012). A mixture distribution of spatial attention. Experimental Psychology, http://dx.doi.org/10.1027/1618-3169/a000182. Funes, M. J., Lupiáñez, J., & Milliken, B. (2007). Separate mechanisms recruited by exogenous and endogenous spatial cues: Evidence from a spatial Stroop paradigm. Journal of Experimental Psychology: Human Perception and Performance, 33, 348–362. Gobell, J. L., Tseng, C., & Sperling, G. (2004). The spatial distribution of visual attention. Vision Research, 44, 1273–1296. Henderson, J. M., & Macquistan, A. D. (1993). The spatial distribution of attention following an exogenous cue. Perception & Psychophysics, 53, 221–230. Hughes, H. C., & Zimba, L. D. (1985). Spatial maps of directed visual attention. Journal of Experimental Psychology: Human Perception and Performance, 11, 409–430. Jans, B., Peters, J. C., & DeWeerd, P. (2010). Visual spatial attention to multiple locations at once: The jury is still out. Psychological Review, 117, 637–684. Kahneman, D. (1973). Attention and effort. Englewood Cliffs, NJ: Prentice Hall. Kahneman, D., Treisman, A., & Gibbs, B. J. (1992). The reviewing of object files: Objectspecific integration of information. Cognitive Psychology, 24, 175–219. Kerzel, D., Born, S., & Schönhammer, J. (2012). Perceptual grouping allows for attention to cover noncontiguous locations and suppress capture from nearby locations. Journal of Experimental Psychology: Human Perception and Performance, 38, 1362–1370. Klein, R. M., Christie, J., & Morris, E. P. (2005). Vector averaging of inhibition of return. Psychonomic Bulletin & Review, 12, 295–300. Klein, R. M., & Shore, D. (2000). Relations among modes of visual orienting. In S. Monsell, & J. Driver (Eds.), Attention & performance XVIII: Control of cognitive processes (pp. 195–208). Cambridge, MA: MIT Press.

13

Kraft, A., Müller, N. G., Hagendorf, H., Schira, M. M., Dick, S., Fendrich, R. M., et al. (2005). Interactions between task difficulty and hemispheric distribution of attended locations: Implications for the splitting attention debate. Brain Research. Cognitive Brain Research, 24, 19–32. Kristjansson, A., Mackeben, M., & Nakayama, K. (2001). Rapid, object-based learning in the deployment of transient attention. Perception, 30, 1375–1387. Lasaponara, S., Chica, A. B., Lecce, F., Lupianez, J., & Doricchi, F. (2011). ERP evidence for selective drop in attentional costs in uncertain environments: Challenging a purely premotor account of covert orienting of attention. Neuropsychologia, 49(9), 2648–2657. Lauwereyns, J. (1998). Exogenous/endogenous control of space-based/object-based attention: Four types of visual selection? European Journal of Cognitive Psychology, 10, 41–74. Luck, S. J., & Vogel, E. K. (1997). The capacity of visual working memory for features and conjunctions. Nature, 390, 279–281. Lupiáñez, J., Milliken, J., Solano, C., Weaver, B., & Tipper, S. (2001). On the endogenous control over the time course of Inhibition Of return. Quarterly Journal of experimental Psychology, 54(A), 753–773. Lupiáñez, J., Mártín-Arévalo, E., & Chica, A. B. (2013). Is Inhibition of Return due to attentional disengagement or to a detection cost? The Detection Cost Theory of IOR. Psicológica, 34, 221–252. Lupiáñez, J., Ruz, M., Funes, M. J., & Milliken, B. (2007). The manifestation of attentional capture: Facilitation or IOR depending on task demands. Psychological Research, 71, 77–91. Macquistan, A. (1997). Object-based allocation of visual attention in response to exogenous, but not endogenous, spatial precues. Psychonomic Bulletin & Review, 4, 512–515. McLachlan, G. J., & Peel, D. (2000). Finite mixture models. New York, NY: Wiley. McMains, S. A., & Somers, D. C. (2005). Processing efficiency of divided spatial attention mechanisms in human visual cortex. Journal of Neuroscience, 25, 9444–9448. Müller, M. M., Malinowski, P., Gruber, T., & Hillyard, S. A. (2003). Sustained division of the attentional spotlight. Nature, 424(6946), 309–312. Müller, H. J., & Rabbitt, P. M. (1989). Reflexive and voluntary orienting of visual attention: Time course of activation and resistance to interruption. Journal of Experimental Psychology: Human Perception and Performance, 15, 315–330. Nakayama, K., & Mackeben, M. (1989). Sustained and transient components of focal visual attention. Vision Research, 29, 1631–1646. Posner, M. I. (1980). Orienting of attention. Quarterly Journal of Experimental Psychology, 32, 3–25. Prinzmetal, W., Presti, D. E., & Posner, M. I. (1986). Does attention affect visual feature integration? Journal of Experimental Psychology: Human Perception and Performance, 12, 361–369. Rizzolatti, G., Riggio, L., Dascola, I., & Umiltá, C. (1987). Reorienting attention across the horizontal and vertical meridians: Evidence in favor of a premotor theory of attention. Neuropsychologia, 25, 30–36. Rizzolatti, G., Riggio, L., & Sheliga, B. M. (1994). Space and selective attention. In C. Umiltà, & M. Moscovitch (Eds.), Attention and performance, XV. (pp. 231–265). Cambridge, MA: MIT Press. Scharlau, I. (2004). Evidence for split foci of attention in a priming paradigm. Perception & Psychophysics, 66, 988–1002. Schneider, K., A., & Bavelier, D. (2003). Components of visual prior entry. Cognitive Psycholology, 47, 333–366. Shore, D. I., Spence, C., & Klein, R. M. (2001). Visual prior entry. Psychological Science, 12, 205–212. Standage, D. I., Trappenberg, T. P., & Klein, R. M. (2005). Modeling divided visual attention with a winner-take-all network. Neural Networks, 18, 620–627. Woodman, G. F., Vecera, S. P., & Luck, S. J. (2003). Perceptual organization influences visual working memory. Psychonomic Bulletin & Review, 10, 80–87.

Spatial working memory capacity predicts bias in estimates of location.

Role of attentional tags in working memory-driven attentional capture.

Relationship between spatial ability, visuospatial working memory and self-assessed spatial orientation ability: a study in older adults.

Visuospatial working memory is severely impaired in Bálint syndrome patients.

Different Cortical Mechanisms for Spatial vs. Feature-Based Attentional Selection in Visual Working Memory.

No fixed item limit in visuospatial working memory.

Visuospatial bootstrapping: implicit binding of verbal working memory to visuospatial representations in children and adults.

Emotional content of stimuli improves visuospatial working memory.

In competition for the attentional template: can multiple items within visual working memory guide attention?

Individual differences in working memory capacity and attentional control.

Video Game Training Enhances Visuospatial Working Memory and Episodic Memory in Older Adults.

Predictive cues and age-related declines in working memory performance.

Multiple gates on working memory.

Schizophrenics show spatial working memory deficits.

A comparison of attentional bias towards drug cues in addicts and non-addicts.

Atomoxetine effects on attentional bias to drug-related cues in cocaine dependent individuals.

Resource-sharing in multiple-component working memory.

Planning and spatial working memory following frontal lobe lesions in man.

Drinkers' memory bias for alcohol picture cues in explicit and implicit memory tasks.

Multiple Strategies for Spatial Integration of 2D Layouts within Working Memory.

Inhibition of Connexin43 Hemichannels Impairs Spatial Short-Term Memory without Affecting Spatial Working Memory.

Attentional bias to food-related visual cues: is there a role in obesity?

Enhanced attentional bias towards sexually explicit cues in individuals with and without compulsive sexual behaviours.

Different neuronal computations of spatial working memory for multiple locations within versus across visual hemifields.