Acta Psychologica 152 (2014) 75–83

Contents lists available at ScienceDirect

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

Attentional cueing by cross-modal congruency produces both facilitation and inhibition on short-term visual recognition☆ Elena Makovac a,b,⁎, Sze Chai Kwok a, Walter Gerbino b a b

Neuroimaging Laboratory, Santa Lucia Foundation, Rome, Italy Psychology Unit, Department of Life Sciences, University of Trieste, Italy

a r t i c l e

i n f o

Article history: Received 4 November 2013 Received in revised form 14 July 2014 Accepted 16 July 2014 Available online xxxx PsycINFO classification: 2346 2343 2320 Keywords: Attention Cross-modal binding Inhibition of return Inverse effectiveness rule Multisensory integration Visual working memory

a b s t r a c t The attentional modulation of performance in a memory task, comparable to the one obtained in a perceptual task, is at the focus of contemporary research. We hypothesized that a biphasic effect (namely, facilitation followed by inhibition) can be obtained in visual working memory when attention is cued towards one item of the memorandum and participants must recognize a delayed probe as being identical to any item of the memorandum. In every trial, a delayed spiky/curvy probe appeared centrally, to be matched with the same-category shape maintained in visual working memory which could be either physically identical (positive trials) or only categorically similar (negative trials). To orient the participant's attention towards a selected portion of a two-item memorandum, a (tzk/wow) sound was played simultaneously with two lateral visual shapes (one spiky and one curved). Our results indicate that an exogenous attentional shift during perception of the memorandum, induced by a congruent audio–visual pairing, first facilitates and then inhibits the recognition of a cued item (but not of a non-cued item) stored in visual working memory. A coherent pattern of individual differences emerged, indicating that the amount of early facilitation in congruent-sound trials was negatively correlated with recognition sensitivity in no-sound trials (suggesting that the inverse effectiveness rule may also apply to memory) and positively correlated with later inhibition, as well as with the self-reported susceptibility to memory failures. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Competition for representational resources makes attention a crucial factor in both perceptual and reflective domains, supporting comparisons and analogies between external and internal attention that are particularly relevant for research on working memory, which lies at their intersection (Chun, Golomb, & Turk-Browne, 2011). Attention affects encoding (Eger, Henson, Driver, & Dolan, 2004) as well as memory retrieval (Guerin, Robbins, Gilmore, & Schacter, 2012; Wolfe, Reinecke, & Brawn, 2006). Selection in the domains of visual perception and visual memory appears to be supported by similar neural mechanisms (Astle, Scerif, Kuo, & Nobre, 2009; Kuo, Rao, Lepsien, & Nobre, 2009; Nobre et al., 2004). Spatial memory is distorted by a task-irrelevant exogenous cue (Van der Stigchel, Merten, Meeter, & Theeuwes, 2007), and object memory deteriorates when attention shifts away from object location during retention (Williams, Pouget, Boucher, & Woodman, 2013).1 ☆ Authors thank David Pearson for comments on an earlier version of the manuscript. Supported by Miur-Cofin 2008 (PRIN200879EB93) and EcoAutobus-Industria (MS01_00006) 2015 grants. ⁎ Corresponding author at: IRCCS Santa Lucia Foundation, Neuroimaging Laboratory, Via Ardeatina 306, Rome, Italy. Tel.: +39 3382305385. 1 For a recent review on how selective attention can operate on perception, memory, and imagery in equivalent ways see Gosling and Astle (2013).

http://dx.doi.org/10.1016/j.actpsy.2014.07.008 0001-6918/© 2014 Elsevier B.V. All rights reserved.

When observers are required to detect a target following the presentation of an exogenous cue that attracts spatial attention towards its location, the manipulation of cue–target onset asynchrony (CTOA) reveals a characteristic biphasic effect of cueing (Posner & Cohen, 1984). Detection is facilitated at short CTOAs (typically, less than 300 ms), while it deteriorates at longer CTOAs, generating an effect labeled as inhibition of return (IOR). Classic IOR – that is, the detection loss for visual targets displayed in a previously attended location after a critical CTOA – can be explained by the reorienting hypothesis, which states that attention is automatically attracted towards the location of the peripheral cue (i.e., a lateral flash), but is subsequently disengaged from that particular location, because of a compensatory mechanism that inhibits the return of attention to previously attended locations, to maximize efficiency of visual search in a normally complex environment (Danziger, Kingstone, & Snyder, 1998; Klein, 2000). For an alternative, motor-based, view of IOR, as well as for its possible occurrence in a perceptual discrimination task, see Taylor and Donnelly (2002). Attention can be attracted towards locations, objects, and features (Carrasco, 2011). Accordingly, IOR is not unique to spatial attention, having been found also when attention is object-based (i.e., oriented to targets that belong to a previously attended object; List & Robertson, 2007) and feature-based (i.e., oriented to targets that possess a previously attended feature; Busse, Katzner, & Treue, 2006). Outside the visual

76

E. Makovac et al. / Acta Psychologica 152 (2014) 75–83

domain, IOR also occurs in auditory and audio–visual domains, provided that a second cue redirects attention back to the center (Reuter-Lorenz & Rosenquist, 1996). Based on the idea that external and internal types of attention may share similar mechanisms, attentional cueing effects have been recently investigated in memory. Besides the well-known facilitatory effect of cueing on encoding (Uncapher, Hutchinson, & Wagner, 2011), attention can improve the maintenance of objects in memory and increase the probability of their recall (Murray, Nobre, Clark, Cravo, & Stokes, 2013). Johnson et al. (2013) manipulated participants' internal attention by presenting two items followed by a cue that required them to selectively think back to (i.e., refresh) only one item. Slower responses for refreshed than unrefreshed items revealed an IOR-like effect within working memory. In this case, internal attention was directed towards the semantic meaning of a word, which was independent of the memory of its spatial location. This again suggests that the memory-based IOR-like effect is not exclusively linked to spatial attention, but can arise from semantic cueing, as described in the perceptual domain (Fuentes, Vivas, & Humphreys, 1999). Cross-modal congruency refers to the correspondence between inputs to different modalities that can make sounds and visual shapes perceptually similar (Köhler, 1929). Following previous demonstrations that cross-modal congruency can improve recognition (Murray et al., 2004), speed up cross-modal event detection (Makovac & Gerbino, 2010), and modulate attention (Chiou & Rich, 2012), here we asked whether the exogenous cueing of a perceptual event (i.e., the automatic orienting of external attention towards one of two visual shapes) can influence recognition. We expected that cross-modal congruency, by promoting multisensory integration and affecting the deployment of attention, would generate a biphasic effect of cueing on recognition performance. In particular, we explored the possibility that visual working memory (VWM) displays a biphasic effect of attentional cueing (facilitation followed by inhibition). Participants in our experiment were shown one spiky and one curvy shape on the left/right of the fixation point and maintained them in VWM until a central probe prompted for an old/new response based on physical identity. A new response (negative trials) was required when the match between the probe and the item of the memorandum with the same contour type was categorically similar (but not identical), whereas an old response was required when an identical probe was presented. Our paradigm included multisensory and unisensory conditions. In multisensory conditions the memorandum was presented together with a simultaneous sound whose auditory features were congruent with the features of one of the two visual shapes, giving rise to cued trials. Cued trials were either valid (if sound and probe were congruent) or invalid (if sound and probe were incongruent). The unisensory condition included uncued trials (neutral; no sound was presented). Importantly, the sound in the multisensory conditions was task-irrelevant, thus making cued and uncued (neutral) trials formally equivalent in terms of task demand. Expected effects of cueing (facilitation followed by inhibition) should be dependent on automatic cross-modal binding. We assumed that: (a) cross-modal congruency promotes the automatic binding of the sound with only one visual shape; (b) attention is exogenously oriented towards the cued shape; (c) multisensory integration enhances encoding (Lehmann & Murray, 2005; Nyberg, Habib, McIntosh, & Tulving, 2000; Wheeler, Petersen, & Buckner, 2000) by driving exogenous attention towards the multisensory event comprising congruent sound/cued shape combination (Spence, McDonald, & Driver, 2004; Talsma, Senkowski, Soto-Faraco, & Woldorff, 2010); and (d) the central probe produces a disengagement of attention from the peripheral cued shape, and gives rise to IOR (Posner & Cohen, 1984). We expected the following two results: 1) Facilitation; when the probe is displayed immediately after the memorandum, recognition in valid trials should be facilitated,

because the exogenously cued shape should benefit from enhanced encoding, at the expense of poorer encoding of the uncued shape; 2) Inhibition; at a longer probe delay (around 1 s), recognition in valid trials is inhibited, as revealed by a reduction in recognition performance for the exogenously cued shape (Lupiáñez, Milán, Tornay, Madrid, & Tudela, 1997; Massen & Stegt, 2007). We also explored the correlation between individual differences in performance and the self-reported frequency of cognitive mistakes, measured by the Memory and Distractibility subscales of the Cognitive Failures Questionnaire (CFQ; Broadbent, Cooper, FitzGerald, & Parkes, 1982; Wallace, Kass, & Stanny, 2002). We hypothesized that the amounts of the expected facilitation and inhibition effects (which should reveal how attention operates in VWM) could be positively correlated with the awareness of the propensity to attention and memory errors in everyday life, as measured by the relevant CFQ items. The reliability and validity of CFQ in quantifying the propensity for making mistakes have been extensively studied (Forster & Lavie, 2007; Kanai, Dong, Bahrami, & Rees, 2011; Martin & Jones, 1983; Tipper & Baylis, 1987). However, CFQ scores might also reflect metacognitive worries (Mecacci & Righi, 2006) and the tendency to pessimistic self-evaluations (van Doorn, Lang, & Weijters, 2010). 2. Method 2.1. Participants Twenty right-handed undergraduates (14 females, mean age = 24 years, age range 19–29, SD = 3.5) participated in the experiment. All participants had normal hearing and normal/corrected-to-normal visual acuity. All gave their prior informed consent, were tested individually, and received course credit. 2.2. Stimuli, apparatus, and procedure As regards visual stimuli, inspired by Köhler (1929, Figs. 18 and 19) we generated 40 unfamiliar shapes — 20 with spiky contours containing abrupt and frequent (median = 28 in the 14–44 range) discontinuities, and 20 with curvy contours containing smooth and infrequent (median = 8 in the 4–12 range) changes of curvature polarity (i.e., from convexity to concavity and vice versa). They were drawn manually, adding variable portions (spiky vs. curvy) to a central disk subtending about 3° and reaching a maximum angular extent of about 9.0° at presentation (Fig. 1A) (see Supplementary material for the whole set of visual shapes). We were aware that spiky and curvy shapes differed in their complexity, if this is defined as a function of the number of contour changes (Baylis & Driver, 2001). The shapes appeared light gray (20.7 cd/m2) on a black screen (5 cd/m2). Nine practice pairs and 288 experimental pairs of visual shapes, to be used as memoranda, were extracted from the 400 spiky–curvy pairs, balancing for position. As regards acoustic stimuli, we used Audacity® 2.0 to generate two 200-ms sounds: tzk, a spiky sound with abrupt and frequent changes of intensity, and wow, a soft sound with smooth and infrequent changes of intensity.2 Participants were seated 57 cm away from a dimly illuminated screen and instructed to pay attention to visual memoranda, treat sounds as task irrelevant, and prioritize accuracy over speed when matching the memorandum to the probe for physical identity. After a 10-min dark adaptation period and a 9-trial practice block, four 72-trial blocks were presented, lasting about 12 min each, with 3-min breaks between blocks. Every trial started with a 1000-ms green fixation cross displayed at the center of the screen, followed by the onset of a 150-ms memorandum made of two shapes (with their centers at ± 7.5° of 2 The audio files “tzk_C.wav” and “wow_C.wav” are available as Supplementary material.

E. Makovac et al. / Acta Psychologica 152 (2014) 75–83

77

Fig. 1. (A) Two spiky and two curvy shapes from a set of 40 shapes used in memorandum pairs and as single probes. (B) Schematic representation and timeline of a positive trial with a spiky probe identical to the left item of the memorandum (correct response = old). Depending on the presence/absence of the sound and on sound–probe congruency, this trial could be neutral (no sound), valid (tzk), or invalid (wow). Audio files with tzk and wow sounds are available as Supplementary material.

horizontal eccentricity from the fixation cross) and of a simultaneous sound, if present. After a 150-ms visual noise mask covering each shape and a 0/1000/6000 ms interval from mask offset, a spiky/curvy probe appeared centrally for a maximum exposure of 2200 ms (Fig. 1B). In half trials the probe was physically identical to one shape of the memorandum (correct response = old); in the other half it was different from both shapes (correct response = new), but always belonging to the experimental set and therefore categorically similar to one shape of the memorandum. Participants pressed either the “1” key for old or the “3” key for new, terminating the probe exposure. A white fixation cross was presented in the intertrial interval, whose duration was pseudo-randomly selected from a truncated exponential distribution favoring short values (range: 816–6868 ms). Responses delivered after the end of the intertrial interval were discarded. Note that this procedure allowed participants to freely inspect the probe until they were ready for responding (given that 2200 ms was a fairly long probe exposure), making response time an index of the amount of information extracted from the probe before its disappearance. Hence, response times (the amount of probe information) reflected the participant's confidence level on delivering an accurate response. We ran the experiment using MATLAB® 7.4 and the Cogent Toolbox (www.vislab.ucl.ac.uk/cogent.php). Visual stimuli appeared on an Acer Aspire 19-inch screen, 60 Hz, 1024 × 768 pixels, set at 50% brightness and 90% contrast. Sounds came from two loudspeakers hidden by the screen, located behind the two items of the memorandum. We conducted a pilot study, with the same set-up in eliciting sounds as in the main experiment. Twenty independent observers completed two trials each (one tzk and one wow), while facing the screen with their eyes closed, judging if a sound came from left, center, or right, relative to the straight-ahead direction. We collected 7 left, 29 center, and 4 right responses, indicating no lateral bias in the absence of visual stimuli. As regards the multisensory trials in which a tzk/wow sound was played together with the presentation of two lateral visual shapes, one spiky and one curvy, we ran a control experiment, briefly reported as Supplementary material, to show that stimulus condition in multisensory trials of the main experiment favored the illusory displacement of the central sound source towards the position of the congruent shape.

Data from the control experiment confirmed our expectation that the congruent item of the visual shape pair could attract the simultaneous central sound (e.g., a central tzk sound was more likely to appear as coming from the left when the visual shape pair was spiky–curvy than curvy–spiky). Hence, we suggest that multisensory trials of our main experiment fulfilled the conditions for a ventriloquist effect, though its actual occurrence in every trial could not be monitored. The possible role of the ventriloquist effect in our paradigm, in the context of the relevant literature, is discussed in Section 4.1.3 The experimental design included three within-subjects factors: Probe (spiky, curvy), Validity (valid, invalid, neutral), and Delay (150, 1150, 6150 ms). Trials were valid when the sound, the cued item of the memorandum, and the probe were categorically congruent (e.g., all spiky); invalid when the sound and the cued item of the memorandum were crossmodally congruent, but the probe was categorically incongruent with both (e.g., tzk cueing the spiky item of the memorandum, followed by a curvy probe that prompted matching it to the uncued item). Performance in valid and invalid trials was compared to baseline performance in nosound neutral trials. Valid trials included two congruent sound–probe couplings (either “tzk ⇒ spiky probe” or “wow ⇒ curvy probe”) and could be positive (probe identical to the cued shape = old) or negative (probe literally different from the cued shape, though belonging to the same category = new). Similarly, invalid trials included two incongruent sound–probe couplings (either “tzk ⇒ curvy probe” or “wow ⇒ spiky probe”) and could also be positive or negative. Since our task probed recognition memory, to evaluate the biphasic effect of cueing we focused on accuracy as the main dependent measure of performance and decided to parameterize it as sensitivity according to Signal Detection Theory (SDT). In our paradigm, response time (RT) included the variable probe exposure interval necessary to the participant for cumulating a sufficient amount of probe information and reaching a reasonably confident old/new choice; for this reason, RTs 3 The Supplementary material also includes a demonstration (“VE_demo.ppsm”), in which a spiky–curvy pair is displayed in alternating frames, accompanied by either a tzk or a wow sound. In such conditions, the tzk sound is more likely to appear as coming from the left and the wow from the right.

78

E. Makovac et al. / Acta Psychologica 152 (2014) 75–83

were taken as a source of evidence on the participant's confidence in matching the probe to the memorandum. Before the experiment every participant was asked to fill out an Italian version of the CFQ, conceived by Broadbent et al. (1982) to assess the tendency to make everyday mistakes.4 The CFQ requires respondents to rate the frequency with which, in the last 6 months, they experienced 25 common cognitive failures in perception, memory, and motor function on a scale from “never” (0) to “very often” (4). Given our experimental paradigm, we focused on the Memory and Distractibility subscales derived from a previous factor analysis (Wallace et al., 2002). Individual CFQ scores on such subscales were correlated with recognition sensitivity at different delays.5

unequal variance model we used RTs to categorize confidence. For every participant, each of the 9 RT distributions of the Validity × Delay design (independent of response correctness) was bisected and responses were associated with a dichotomous confidence level (low for RT N median, high for RT ≤ median). This procedure allowed us to obtain three points for each ROC line and to estimate the following unequal variance parameters: the sensitivity measure d a (whose meaning is equivalent to the meaning of d′ for the equal variance model), three ca values (one for each of the three response criteria derived from the assumption that short/long RT durations were indicative of low/high confidence), and the SD ratio of old/new probe distributions. The distributions of da , ca , and SD ratio values are discussed in Sections 3.2–4, respectively.

2.3. Data analysis 3. Results There were 16 trials (8 old and 8 new probes) in each of the 18 conditions of the 3-factor design. The values of SDT indices were derived from old/new responses in the following way. Less than 1% trials were not analyzed because the response was too fast (RT b 200 ms), too slow (RT N 3 standard deviations above the individual mean), or missing. In positive trials old responses were classified as Hits (remembered) and new responses as Misses (forgotten); in negative trials old responses were classified as False Alarms and new responses as Correct Rejections. Since the literature on the SDT approach to recognition shows that the standard deviation of old items tends to be larger than the standard deviation of lures (Ratcliff & Starns, 2009), to recover the ratio of standard deviations (SD ratio) of old items' relative lures we categorized RTs as short/long to derive an index of response confidence and to analyze recognition performance according to the unequal variance model (Macmillan & Creelman, 2005).6 First, we analyzed RTs for the Probe × Validity × Delay design, to control for the effects of instructions, which stressed accuracy and not speed, and to evaluate unplanned differences between spiky vs. curvy probes, computing individual means on the basis of a minimum of 14 RTs out of 16 (first part of Section 3.1). To evaluate a possible speed–accuracy trade-off we also inspected the correlation between 1/RT and d′ values (second part of Section 3.1). Then, to apply the

4 Our Italian version of the CFQ differed from the Italian version validated by Stratta, Rinaldi, Daneluzzo, and Rossi (2006) in minor linguistic details. The most relevant difference was our usage of the second person singular (tu), instead of the third person singular (lei) to address the respondent. This choice was motivated by age proximity between the experimenter (EM) and participants, which would have made the usage of the third person singular (correct but formal, in Italian) quite unnatural. 5 The Memory subscale of the CFQ includes items like “Do you find you forget whether you've turned off a light or a fire or locked the door?” and “Do you find you forget appointments?”. The Distractibility subscale of the CFQ includes items like “Do you read something and find you haven't been thinking about it and must read it again?” and “Do you have trouble making up your mind?”. 6 Data from an old/new recognition task fit into the framework of SDT naturally (Macmillan & Creelman, 2005). One pair of Hit (H) and False Alarm (FA) proportions allows to estimate recognition sensitivity (i.e., the accuracy of old/new judgments), along the familiarity continuum, as well as response bias (i.e., the observer's propensity for the old vs. new response). In the equal-variance SDT model sensitivity is represented by d′, the separation between the means of two Gaussian distributions with the same standard deviation, one for old (signal + noise) and one for new (noise) probes; while bias is represented by the position of the criterion c along the familiarity continuum. The two indices are related to raw measures of accuracy and bias (difference score and yes rate, respectively): d′ = z(H) − z(FA); c = −0.5 [(z(H) + z(FA)]. To apply the unequal variance SDT model one needs more pairs of H and FA proportions, one for each response criterion associated to a different degree of observer's confidence in his/her old/new responses, under the same level of sensitivity. In the unequal variance model the appropriate indices of sensitivity and bias are da and ca, respectively; formulas for their computation are similar to those for d′ and c, taking into account of the inequality of standard deviations. To estimate the SD ratio, [z(FA), z(H)] points are plotted in binormal coordinates to interpolate a ROC (receiver operating characteristic) straight line that fully describes performance in a given condition. The assumption of equal variance (which underlies the use of d′ and c indices) would have been inadequate in our case, as shown by SD ratio data reported in Section 3.4. However, we ran all analyses also according to the equal variance model and found convergent results for sensitivity and criterion measures. The outcomes of ANOVAs on d′ and c values are not reported but are available from authors.

3.1. RTs Fig. 2 shows mean RTs and s.e.m. in different conditions. Since exploratory 3-way ANOVAs on RT and transformed 1/RT values were consistent, only the outcome for transformed data is reported here. Responses were faster for curvy than spiky probes, 908 vs. 935 ms, F(1, 19) = 7.34, p b .05, η2p= .28. The main effect of Delay was also significant, F(2, 38) = 27.79, p b .001, η2p= .73, due to longer RTs at Delay6150 (980 ms) than at Delay150 (904 ms) and Delay1150 (881 ms). Neither the main effect of Validity, F(2, 38) = 1.83, p = .17, η2p= .32, nor the 2-way interactions were significant: Probe × Validity, F(2, 38) = 1.62, p = .21, η2p = .16; Probe × Delay, F(2, 38) = 1.85, p = .17, η2p= .09; Validity × Delay, F b 1. The 3-way interaction was significant, F(4, 76) = 2.97, p b .05, η2p=.43. Before interpreting RT effects we checked the possible existence of a speed/accuracy trade-off among the 18 conditions of our Probe (2) × Validity (3) × Delay (3) design. For all 18 conditions we computed the mean 1/RT and d′ values for every participant individually. Contrary to a speed– accuracy trade-off, the correlation between 1/RT and d′ values, r = .33 on average, was positive for 17 out of 20 participants, p b .01 (two-tailed sign test). This result ruled out the possibility that performance in our paradigm was affected by a global compensatory strategy (slowing down the response to cumulate more probe information and achieve better accuracy). Rather, the RT variability reflected basic differences in task difficulty among our 18 conditions, which we interpreted as follows. First, participants needed more time to cumulate enough information from spiky than curvy probes, consistent with figural complexity: spiky contours contained more changes of direction than curvy contours (Baylis & Driver, 2001); spiky shapes had more pointing directions than curvy shapes, involving higher directional ambiguity (Palmer & Bucher, 1981). Second, the longest target-probe delay was associated with a clear slowing down of response speed, reflecting the need to acquire more probe information to perform the required matching. However, mean RTs were not a linear function of delay. The fastest responses were obtained with the intermediate delay, which probably was long enough to prevent information transferral to VWM but not so long to induce a loss of confidence in performing a successful match. The 3-way interaction in RTs suggests that all three factors probably affected the behavioral pattern; these effects are further examined using the sensitivity measure (see analysis on da values below). The obtained effects on RT variability across experimental conditions should not obscure the within-condition RT variability that we planned to use as information about confidence, to derive SDT unequal-variance indices as reported in the following sections. 3.2. da values Having ruled out a speed–accuracy trade-off, to better evaluate the presence of a biphasic effect of cueing (which should be revealed by a significant Validity × Delay interaction on da values) we used long/short RTs

E. Makovac et al. / Acta Psychologica 152 (2014) 75–83

79

Fig. 2. Mean RTs and s.e.m. in the 18 conditions of the Probe × Validity × Delay design. Recognition was faster for curvy than spiky probes. With both probes participants were much slower when the probe appeared 6150 ms after the offset of the memorandum, relative to shorter delays.

as evidence of low/high confidence and analyzed data according to the unequal-variance SDT model, taking da as the most appropriate measure of recognition sensitivity.7 In neutral trials, when attention was evenly distributed over the whole memorandum, d a values at Delay 150 and Delay1150 did not differ, 1.77 vs. 1.88, t(19) = 1.24, two-tailed, p = .23, d = .27, while a significant deterioration of VWM information occurred at Delay6150, da = 1.13, t(19) = 6.02 and 8.87, twotailed, p b .001, d = 1.34 and 1.72, with respect to Delay 150 and Delay1150. To discount the mere effect of Delay on sensitivity, revealed by performance in neutral trials, and to analyze the pure biphasic effect of attentional cueing, we took sensitivity in neutral trials as a common reference and computed the amount of cueing effect, Δda = [da(valid, invalid) − da(neutral)], as a function of log(Delay). A 2-way ANOVA on Δda values revealed no main effect of Validity, F(1, 19) = 1.28, p = .27, η2p=.06, while the main effect of Delay, F(2, 38) = 4.61, p b .05, η2p=.39, and the Validity × Delay interaction, F(2, 38) = 3.95, p b .05, η2p=.62, was significant (Fig. 3). This interaction was expected, as a consequence of the presence of a biphasic effect of attentional cueing in valid, but not invalid, trials. The biphasic effect of cueing consisted in the combination of facilitation in valid over neutral trials at Delay150, da = 1.99 vs. 1.76, t(19) = 2.73, one-tailed, p b .05, d = .47, followed by inhibition at Delay1150, where Da(valid) = 1.60 was significantly lower than D a(neutral) = 1.88: t(19) = 3.60, one-tailed, p b .01, d = .72, and no evidence of inhibition at Delay 6150 , where d a(valid) was at the baseline. On the contrary, relative sensitivity in invalid trials reduced over time, supported by a lack of significance of Δda at Delay150, t b 1, combined with a significant negative Δda when Delay1150 and Delay6150 were pooled, t(39) = 2.54, one-tailed, p b .05, d = .26. In other words, recognition accuracy monotonically decayed as a function of time in neutral and invalid trials, while it was non-monotonically modulated by exogenous attentional cueing in valid trials.

3.3. ca values As regards response bias, Fig. 4 shows the distribution of the three ca values for the unequal variance model.8 The ANOVA on ca values for the Criterion (3) × Validity (3) × Delay (3) design revealed a main effect of Criterion, F(2, 38) = 16.35, p b .001, η2p=.99, intrinsic to the procedure, a main effect of Validity, F(2, 38) = 7.40, p b .005, η2p=.35, dependent 7

We checked that da values were highly correlated with d′, r = .84. We also computed the c values for the equal-variance model. Correlations between c, mean ca, and median ca values were all very high, r N .96, justifying the use of mean ca as our best estimate of average response bias in each condition of the basic Validity × Delay design. As for c, the propensity towards the old response leads to negative ca values while the propensity towards the new response leads to positive ca values (the unbiased observer being revealed by ca = 0). 8

on the lower frequency of old responses in neutral and invalid trials, ca(neutral + invalid) = .09 vs. ca(valid) = − .02. The main effect of Delay, F(2, 38) = 76.00, p b .001, η2p = .85, was consistent with the conservative drift from Delay150, ca = − .13 (correspondent to a yes rate of about 59.4%), through Delay1150, ca = .07 (yes rate = 46.5%), to Delay6150, ca = .41 (yes rate = 33.3%). Both 2-way interactions involving Delay were also significant. The Validity × Delay interaction, F(4, 76) = 5.50, p b .001, η2p= .57, showed that the increase of ca as a function of Delay was steeper for neutral and invalid trials (Fig. 4A). The Criterion × Delay interaction, F(4, 76) = 7.59, p b .001, η2p= .60, depended on the steepest slope of the median criterion (Fig. 4B). The Criterion × Validity, F(4, 76) = 1.65, p b .17, η2p= .40, and the 3-way interaction, F b 1, were not significant. 3.4. SD ratio We also analyzed the SD ratio of old/new probe distributions (Fig. 5).9 The average SD ratio was 1.37, quite close to values found in the recognition literature (Ratcliff & Starns, 2009). Together with the high correlations between sensitivity (d′, da) and criterion (c, ca) measures, this correspondence corroborated the validity of our implementation of the unequal variance model. A 2-way ANOVA on SD ratios for the Validity × Delay design showed that the main effect of Delay was significant, F(2, 38) = 4.46, p b .01, η2p=.19, while the main effect of Validity, F b 1, and the 2-way interaction, F(4, 76) = 1.18, p = .33, η2p=.23, were not. The main effect of Delay was consistent with the SD ratio getting closer to 1 as memory becomes weaker (Ratcliff & Starns, 2009, p. 80). Despite the lack of significance of the 2-way interaction, evidence of IOR prompted us to test whether the SD ratio reduction between Delay150 and Delay1150 was significantly larger in valid vs. neutral trials. The difference approached statistical significance, .505 vs. .163, t(19) = 2.09; one-tailed, p = .051, d = .47, consistent with the hypothesis that about 1 s after the removal of the memorandum attention was disengaged from the cued item, producing a disproportionally larger VWM loss in valid trials, relative to invalid trials. 3.5. Individual differences and CFQ scores As regards interindividual differences, the negative correlation between participants' da and mean ca values in the 9 Validity × Delay conditions was significant, r = −.39, t(178) = −5.73, two-tailed, p b .001, supporting the idea that participants who are at least partially aware of their poor recognition performance tend to be more conservative. In valid trials, facilitation (Δda at Delay150) positively correlated with IOR 9 ROC lines in binormal coordinates, for each of the 9 conditions of the Validity (3) × Delay (3) design, are plotted in Fig. S2, available as Supplementary material.

80

E. Makovac et al. / Acta Psychologica 152 (2014) 75–83

Fig. 5. Ratio of standard deviations (SD ratio) for old/new probes according to the unequal variance SDT model, as a function of log(Delay). As reported in the recognition literature, the SD ratio was always larger than 1. The SD ratio decreased as the delay increased, with a disproportionally large decrement in valid trials. Fig. 3. Relative sensitivity Δda (i.e., deviation from the corresponding da value in neutral trials) in valid and invalid trials, as a function of log(Delay). Sensitivity in valid trials was better than the baseline when the probe was displayed immediately after the memorandum, while it dropped below the baseline and the level of invalid trials at the 1150-ms delay, supporting the hypothesis of an IOR-like effect in recognition memory. Sensitivity in invalid trials was always worse than in corresponding neutral trials and slowly decayed over time.

Fig. 4. Mean ca values as a function of log(Delay) for the three Validity levels (panel A) and for the three Criterion levels (panel B). Error bars (±1 s.e.m.) are sometimes smaller than symbols. In general, the propensity for the old response decreased as the probe delay increased from 150 to 6150 ms. Panel A shows that this tendency was weaker in valid than neutral and invalid trials. Panel B shows that the relative positions of the three criteria gradually changed as a function of delay.

loss (the difference between da at Delay150 and da at Delay1150), r = .58, t(18) = 3.00, two-tailed, p b .01, suggesting that participants who better integrated the sound with the congruent shape of the memorandum at encoding also exhibited a larger IOR loss. Two correlations emerged at Delay150. Facilitation in valid trials and sensitivity in neutral trials were negatively correlated, r = −.37, t(18) = 1.70, one-tailed, p = .053, consistent with an interindividual version of the inverse effectiveness rule of multisensory integration (Stein & Meredith, 1993), showing that participants with a relatively poor performance in neutral unisensory trials show a higher facilitation in valid multisensory trials. Facilitation in valid trials positively correlated with the mean c a value, r = .49, t(18) = 2.38, two-tailed, p b .05, indicating that more conservative participants enjoyed a larger facilitation. No correlations were found in the analogous measures at Delay1150 and Delay6150. Following Wallace et al. (2002), we computed individual CFQ-scores for the Memory and Distractibility subscales, and correlated such scores with the individual amounts of early facilitation and late inhibition, with the expectation that participants who subjectively reported a higher proportion of cognitive failures related to memory and/or distractibility were more prone to the biphasic effect of cueing, exhibiting larger amounts of both early facilitation and late inhibition. As expected, the CFQ-Memory score was positively correlated with facilitation, r = .38, t(18) = 1.75, and inhibition, r = .41, t(18) = 1.89 (both one-tailed, p b .05). The correlations between the CFQ-Distractibility score and facilitation (r = .13, t b 1) and between the CFQ-Distractibility score and inhibition (r = .09, t b 1) were positive but not statistically significant. In other words, recognition performance was related to the self-evaluated capability of holding information in VWM and not to the self-evaluated mental concentration ability (the construct behind the Distractibility subscale).10 To evaluate possible causes of the inhibitory effect at Delay1150, we ran a multiple regression analysis with IOR loss as the dependent variable, and CFQ-Memory and facilitation as predictors. The overall portion of explained variance was significant, r2 = .37, F(2, 17) = 5.09, p b .05, with a substantial effect of facilitation, b = 0.500, t(17) = 2.38, one tailed, p b .05, and a negligible contribution of CFQ-Memory, b = 0.006, t(17) = 1.06.

10 CFQ validation studies agree in attributing a larger proportion of explained variance to the Memory factor than to the Distractibility factor (Stratta et al., 2006; Wallace et al., 2002).

E. Makovac et al. / Acta Psychologica 152 (2014) 75–83

4. Discussion 4.1. A biphasic effect of multisensory cueing in memory We demonstrated a biphasic effect of the multisensory exogenous cueing of attention in a recognition paradigm. A memorandum made of two abstract shapes, one spiky and one curvy, was presented either alone (neutral trials) or together with a simultaneous sound congruent with one shape (valid/invalid trials). After a variable delay, a spiky/curvy probe appeared centrally, either physically identical (positive trials) or only categorically similar (negative trials) to one of the two shapes maintained in VWM. In multisensory trials the probe could belong to the same category as either the cued (valid trials) or uncued (invalid trials) shape. In valid trials cross-modal cueing produced a facilitation when the probe appeared 150 ms after the memorandum and a recognition loss when it appeared after 1150 ms. The whole pattern of recognition sensitivities in neutral, valid, and invalid trials was consistent with an automatic attention shift towards the cued shape, leading to better encoding of such a shape and to poorer encoding of the uncued shape, during the presentation of the memorandum and immediately after its removal, followed by a temporary disengagement of attention from the memory representation of the cued shape. The attention shift was mediated by the cross-modal congruency of sound, target shape, and probe. Our results parallel previously reported effects of cross-modal congruency in perceptual tasks (Makovac & Gerbino, 2010; Parise & Spence, 2009), support strong links between attention and VWM (Awh & Jonides, 2001), and are fully consistent with the continuity across external and internal attention (Chun et al., 2011). At Delay150 recognition sensitivity improved in valid but not invalid trials, with respect to neutral trials. Since the facilitation was selective for the cued shape of the memorandum, it could not be attributed to generic alerting. Rather, cross-modal congruency supported multisensory integration, drove attention, and improved the encoding of the congruent shape, consistent with other studies demonstrating that the exogenous cueing of attention determines which information is preferentially stored in VWM (Bays & Husain, 2008). Admittedly, the acoustic cue could have either oriented spatial attention (captured by a ventriloquized multisensory event), or activated a feature-based attentional mechanism (guided by crossmodal congruency within an abstract “feature space”), or produced a combination of both effects. Let us consider the two possible mechanistic explanations in more detail. According to the first, spatial-based mechanism, the successful binding of audiovisual stimuli might result in the ventriloquist effect (Alais & Burr, 2004; Bertelson, 1999; Bertelson, Vroomen, de Gelder, & Driver, 2000; Spence et al., 2004), which consists in the misplacement of the auditory source towards the location of a congruent visual event.11 As a consequence, attention should be attracted towards the ventriloquized item of the memorandum because of the higher salience of a multisensory event (resulting from the successful binding of the sound and the cued visual shape) over a unisensory stimulus (the uncued visual shape). The attentional deployment of spatial attention 11 A demonstration of the effect that our observers might have experienced in multisensory trials, during the simultaneous presentation of the memorandum and the sound, is provided in the Supplementary material (file “VE_demo.ppsm”). Data from the “control experiment” reported as Supplementary material show that the apparent sound source can be spatially captured by the congruent item of a pair of visual shapes. Stimulus conditions in the control experiment were similar to those of our main experiment and to anecdotal occurrences of the ventriloquist effect in everyday TV experience, where the voice is captured by the most congruent character, when two or more characters are on screen and acoustic stimulation comes from a mono channel. The control experiment shows that a ventriloquist effect can occur when two shapes are presented simultaneously with the sound and congruent visual information is provided by static contour features (instead of synchronous lip movements and contextual knowledge). This variant of ventriloquist capture is complementary to the effect studied by Driver (1996), in which two sound streams originate from the same loudspeaker and lip movements of a single actress capture the congruent one.

81

following the ventriloquist effect has been described by Vroomen, Bertelson, & de Gelder (2001) and can be taken as an automatic consequence of audiovisual binding. According to the second, feature-based mechanism, our experimental paradigm involved the covariation of spatial and featural cueing: the two visual shapes of the memorandum differed in both left/right location and spiky/curvy contour features (i.e., if the spiky shape was on the left then the curvy shape was on the right, and vice versa). The simultaneous acoustic cue in valid/invalid trials was central (i.e., equidistant from both items of the memorandum) but always more similar (i.e., featurally closer) to one of the two. Therefore, featural similarity might favor the cross-modal binding of the sound with only one visual shape, despite its spatio-temporally balanced position that would otherwise glue together any sound and both visual shapes with equal strengths. In other terms, the misplacement of the sound source in the environmental space surrounding the observer – though supported by evidence in the literature and our demo in Supplementary material – is not an indispensable component of our paradigm, which only requires an unbalanced binding during the simultaneous presentation of memorandum and sound. An encoding unbalance between the two items of the memorandum might have occurred even without the ventriloquist effect (whose occurrence was likely but impossible to confirm, at least in all trials), as a consequence of a feature-based attentional mechanism elicited by cross-modal congruency. Note that some authors failed to disentangle location-based and object-based mechanisms of IOR in a perceptual task (McAuliffe, Pratt, & O'Donnell, 2001), while others have suggested that location-based, object-based, and feature-based inhibitory effects may combine in an additive fashion (Hotta, Oba, & Ishii, 2010; Leek, Reppa, & Tipper, 2003). Interestingly, facilitation in valid multisensory trials was stronger for participants whose recognition sensitivity was low in neutral unisensory trials, providing evidence that the inverse effectiveness rule of multisensory integration, originally described for neural and behavioral measures of detection (Stein, 2012; Stein, Huneycutt, & Meredith, 1988; Stein & Meredith, 1993), may also hold at the interindividual level in a memory task. 4.2. IOR in VWM At Delay1150 recognition sensitivity was impaired in valid but not invalid trials, showing an IOR-like interference with the maintenance and retrieval of an efficiently encoded shape. The selectivity of the recognition loss for valid trials obtained at Delay1150 rules out the possible dependence of the impaired matching on the temporary unavailability of the probe (due, for instance, to saccadic suppression), since this should also affect performance in invalid trials. However, the inhibitory effect obtained in our experiment differs from classic IOR in several ways. First, the suppression of memory retrieval is strictly internal, being dependent on attention operating within a memory representation. Inhibition may be caused by a misguided location search within visual memory, or alternatively, by an impaired access to one part of it. While previous studies (Nobre et al., 2004) elegantly explored the comparable effects of endogenous cues on attention to external stimuli and memory representations, we manipulated the cue/probe asynchrony to monitor the evolution of cueing effects in memory. Notably, our work provides important implications for the study by Johnson et al. (2013), who reported slower responses towards pictures/ words that were mentally refreshed by internal reflective attention, after participants had encoded the items. In their study reflective attention was controlled during maintenance, whereas we manipulated attention during memorandum exposure (by a congruent cue in another modality) and at retrieval, when the probe acted as a back-to-center cue in valid and invalid (but not neutral) trials. As regards the IOR refractory period, performance was restored after 6 s in our experiment, compared to 20 min in Johnson et al. (2013). This is consistent with

82

E. Makovac et al. / Acta Psychologica 152 (2014) 75–83

the idea that the deployment of internal attention involves an IOR-like effect for previously attended items which is analogous, but not identical, to the IOR effect induced by cueing external attention in the perceptual domain. Previous studies have reported the independence of exogenous and endogenous attentional cueing in a variety of tasks (Berger, Henik, & Rafal, 2005). We suggest that, in our paradigm, the endogenous cueing by the categorical identity of the central probe (spiky vs. curvy) did not prevent the occurrence of an IOR-like effect due to the combined action of the exogenous multisensory cue (during presentation of the memorandum) and the exogenous back-to-center cue represented by probe onset (irrespective of its categorical identity). This inhibition might be linked to the mechanism proposed by Theeuwes, Van der Stigchel, and Olivers (2006), who found that when attention is engaged at a location for later memory recall, oculomotor suppression occurs by deviating saccadic trajectories away from that location, as well as by slowing saccadic movements to the memorized cued location (Belopolsky & Theeuwes, 2011). One could think that this mechanistic explanation holds for the spatial-based account of IOR but not for the featurebased account of IOR (for a debate on this topic see Mazer, 2011). To reconcile the two accounts Hotta et al. (2010) assumed that feature-based inhibitory signals follow the computation of the spatial saliency map. Whether such a saliency map could be applicable also to the memory domain is still unknown, and as such, the explanation of our data in light of their model is speculative. 4.3. Effects on response criterion A modest but significant negative correlation was observed between sensitivity and criterion measures, in line with previous evidence of the association between poorer sensitivity and a more conservative bias (Aberg & Herzog, 2012), and with the conservative drift as the delay increased from 150 to 6150 ms (Fig. 4B). Independent of cue validity, the subjective memory strength decreased as the delay increased, reducing the propensity for the old response. Such correlations co-existed with the overall tendency for participants to be unbiased in valid trials, independent of sensitivity, instead of being slightly conservative in invalid and neutral trials. This suggests that valid trials were taken as conditions of relatively higher evidence, compared to neutral and invalid trials in which the probe should be matched to a previously uncued shape. Our criterion effects in recognition can be related to analogous effects in perception. Like Rahnev et al. (2011) who reported a shift towards a more conservative criterion when the target was less visible, we found that the propensity for the old response decreased at longer delays, when the memory trace was weaker. However, we also found a less conservative criterion in valid than invalid trials. In general, our measurements of sensitivity and criterion support the partial dissociation between objective and subjective components of memory recently studied by Bona, Cattaneo, Vecchi, Soto, and Silvanto (2013). Our participants were implicitly aware that task difficulty increased with delay length and changed their response criterion accordingly, but were unaware of the specific modulation of facilitatory and inhibitory effects by trial validity. 4.4. CFQ, facilitation, and IOR Our analysis of individual differences included the self-reported susceptibility to memory failures, assessed by the Memory subscale of CFQ. We argue that participants who reported a higher frequency of memory failures in daily life obtained a larger benefit from the exogenous cueing of attention in immediate recognition (stronger facilitation in immediate recognition) because they were more prone to cueing by a congruent sound. Early facilitation and later IOR loss were also significantly correlated. However, a multiple regression analysis showed that the interindividual variability of the IOR loss was totally explained by early facilitation, without any significant contribution by the self-reported

susceptibility to memory failures. In other words, we found no evidence that the positive correlation between facilitation in immediate recognition and subsequent loss in delayed recognition – which represents an important outcome of our experiment – did follow from the dependence of both effects upon a generic propensity to memory failures. Rather, such a correlation appears to be a specific feature of the biphasic effect of exogenous attention capture involved in our VWM paradigm.

Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.actpsy.2014.07.008.

References Aberg, K. C., & Herzog, M. H. (2012). Different types of feedback change decision criterion and sensitivity differently in perceptual learning. Journal of Vision, 12, 1–11. Alais, D., & Burr, D. (2004). No direction-specific bimodal facilitation for audiovisual motion detection. Brain Research. Cognitive Brain Research, 19, 185–194. Astle, D. E., Scerif, G., Kuo, B. C., & Nobre, A. C. (2009). Spatial selection of features within perceived and remembered objects. Frontiers in Human Neuroscience, 3, 6. Awh, E., & Jonides, J. (2001). Overlapping mechanisms of attention and spatial working memory. Trends in Cognitive Science, 5, 119–126. Bays, P. M., & Husain, M. (2008). Dynamic shifts of limited working memory resources in human vision. Science, 321, 851–854. Baylis, G. C., & Driver, J. (2001). Shape-coding in IT cells generalizes over contrast and mirror reversal, but not figure-ground reversal. Nature Neuroscience, 9, 937–942. Belopolsky, A. V., & Theeuwes, J. (2011). Selection within visual memory representations activates the oculomotor system. Neuropsychologia, 49, 1605–1610. Berger, A., Henik, A., & Rafal, R. (2005). Competition between endogenous and exogenous orienting of visual attention. Journal of Experimental Psychology: General, 134, 207–221. Bertelson, P. (1999). Ventriloquism: A case of cross-modal perceptual grouping. In G. Aschersleben, T. Bachmann, & J. Müsseler (Eds.), Cognitive contributions to the perception of spatial and temporal events (pp. 347–362). Amsterdam: Elsevier. Bertelson, P., Vroomen, J., de Gelder, B., & Driver, J. (2000). The ventriloquist effect does not depend on the direction of deliberate visual attention. Perception & Psychophysics, 62, 321–332. Bona, S., Cattaneo, Z., Vecchi, T., Soto, D., & Silvanto, J. (2013). Metacognition of visual short-term memory: Dissociation between objective and subjective components of VSWM. Frontiers in Psychology, 4 (article 62). Broadbent, D. E., Cooper, P. F., FitzGerald, P., & Parkes, K. R. (1982). The Cognitive Failures Questionnaire (CFQ) and its correlates. British Journal of Clinical Psychology, 21, 1–16. Busse, L., Katzner, S., & Treue, S. (2006). Spatial and feature-based effects of exogenous cueing on visual motion processing. Vision Research, 46, 2019–2027. Carrasco, M. (2011). Visual attention: The past 25 years. Vision Research, 51, 1484–1525. Chiou, R., & Rich, A. N. (2012). Cross-modality correspondence between pitch and spatial location modulates attentional orienting. Perception, 41, 339–353. Chun, M. M., Golomb, J.D., & Turk-Browne, N.B. (2011). A taxonomy of external and internal attention. Annual Review of Psychology, 62, 73–101. Danziger, S., Kingstone, A., & Snyder, J. J. (1998). Inhibition of return to successively stimulated locations in a sequential visual search paradigm. Journal of Experimental Psychology: Human Perception and Performance, 24, 1467–1475. Driver, J. (1996). Enhancement of selective listening by illusory mislocation of speech sounds due to lip-reading. Nature, 381, 66–68. Eger, E., Henson, R. N., Driver, J., & Dolan, R. J. (2004). BOLD repetition decreases in objectresponsive ventral visual areas depend on spatial attention. Journal of Neurophysiology, 92, 1241–1247. Forster, S., & Lavie, N. (2007). High perceptual load makes everybody equal: Eliminating individual differences in distractibility with load. Psychological Science, 18, 377–381. Fuentes, L. J., Vivas, A. B., & Humphreys, G. W. (1999). Inhibitory mechanisms of attentional networks: Spatial and semantic inhibitory processing. Journal of Experimental Psychology: Human Perception and Performance, 25, 1114–1126. Gosling, S. G., & Astle, D. A. (2013). Directing spatial attention to locations within remembered and imagined mental representations. Frontiers in Human Neuroscience, 7 (Article 154). Guerin, S. A., Robbins, C. A., Gilmore, A. W., & Schacter, D. L. (2012). Interactions between visual attention and episodic retrieval: Dissociable contributions of parietal regions during gist-based false recognition. Neuron, 75, 1122–1134. Hotta, S., Oba, S., & Ishii, S. (2010). Visual attention model involving feature-based inhibition of return. Artificial Life and Robotics, 15, 129–132. Johnson, M. R., Higgins, J. A., Norman, K. A., Sederberg, P. B., Smith, T. A., & Johnson, M. K. (2013). Foraging for thought: An inhibition-of-return-like effect resulting from directing attention within working memory. Psychological Science, 24, 1104–1112. Kanai, R., Dong, M. Y., Bahrami, B., & Rees, G. (2011). Distractibility in daily life is reflected in the structure and function of human parietal cortex. The Journal of Neuroscience, 31, 6620–6626. Klein, R. (2000). Inhibition of return. Trends in Cognitive Sciences, 4, 138–147. Köhler, W. (1929). Gestalt psychology. New York: Liveright.

E. Makovac et al. / Acta Psychologica 152 (2014) 75–83 Kuo, B. C., Rao, A., Lepsien, J., & Nobre, A.C. (2009). Searching for targets within the spatial layout of visual short-term memory. The Journal of Neuroscience, 29, 8032–8038. Leek, E., Reppa, L., & Tipper, S. (2003). Inhibition of return for objects and locations in static displays. Perception & Psychophysics, 65, 388–395. Lehmann, S., & Murray, M. M. (2005). The role of multisensory memories in unisensory object discrimination. Cognitive Brain Research, 24, 326–334. List, A., & Robertson, L. C. (2007). Inhibition of return and object-based attentional selection. Journal of Experimental Psychology: Human Perception and Performance, 33, 1322–1334. Lupiáñez, J., Milán, E. G., Tornay, F. J., Madrid, E., & Tudela, P. (1997). Does inhibition of return occur in discrimination tasks? Yes, it does, but later. Perception & Psychophysics, 59, 1241–1253. Macmillan, N., & Creelman, C. D. (2005). Detection theory: A user's guide. New York: Cambridge University Press. Makovac, E., & Gerbino, W. (2010). Sound-shape congruency affects the multisensory response enhancement. Visual Cognition, 18, 133–137. Martin, M., & Jones, G. V. (1983). Distribution of attention in cognitive failure. Human Learning, 2, 221–226. Massen, C., & Stegt, S. (2007). Inhibition of return impairs episodic memory access. Quarterly Journal of Experimental Psychology, 60, 696–707. Mazer, J. A. (2011). Spatial attention, feature-based attention, and saccades: Three sides of one coin? Biological Psychiatry, 69, 1147–1152. McAuliffe, J., Pratt, J., & O'Donnell, C. (2001). Examining location-based and object-based components of inhibition of return in static displays. Perception & Psychophysics, 63, 1072–1082. Mecacci, L., & Righi, S. (2006). Cognitive failures, metacognitive beliefs and aging. Personality and Individual Differences, 40, 1453–1459. Murray, M. M., Michel, C. M., Grave de Peralta, R., Ortigue, S., Brunet, D., Gonzalez Andino, S., et al. (2004). Rapid discrimination of visual and multisensory memories revealed by electrical neuroimaging. NeuroImage, 21, 125–135. Murray, A.M., Nobre, A.C., Clark, I. A., Cravo, A.M., & Stokes, M. G. (2013). Attention restores discrete items to visual short-term memory. Psychological Science, 24, 550–556. Nobre, A.C., Coull, J. T., Maquet, P., Frith, C. D., Vandenberghe, R., & Mesulam, M. M. (2004). Orienting attention to locations in perceptual versus mental representations. Journal of Cognitive Neuroscience, 16, 363–373. Nyberg, L., Habib, R., McIntosh, A.R., & Tulving, E. (2000). Reactivation of encoding-related brain activity during memory retrieval. Proceedings of the National Academy of Sciences of the United States of America, 97, 11120–11124. Palmer, S. E., & Bucher, N. M. (1981). Configural effects in perceived pointing of ambiguous triangles. Journal of Experimental Psychology: Human Perception and Performance, 7, 88–114. Parise, C. V., & Spence, C. (2009). ‘When birds of a feather flock together’: Synesthetic correspondences modulate audiovisual integration in non-synesthetes. PLoS ONE, 4, e5664. Posner, M. I., & Cohen, Y. (1984). Components of visual orienting. In H. Bouma, & D.G. Bouwhuis (Eds.), Attention and performance X: Control of language processes (pp. 531–556). Hillsdale: Erlbaum. Rahnev, D., Maniscalco, B., Graves, T., Huang, E., de Lange, F. P., & Lau, H. (2011). Attention induces conservative subjective biases in visual perception. Nature Neuroscience, 14, 1513–1515.

83

Ratcliff, R., & Starns, J. J. (2009). Modeling confidence and response time in recognition memory. Psychological Review, 116, 59–83. Reuter-Lorenz, P. A., & Rosenquist, J. N. (1996). Auditory cues and inhibition of return: The importance of oculomotor activation. Experimental Brain Research, 112, 119–126. Spence, C., McDonald, J., & Driver, J. (2004). Exogenous spatial-cuing studies of human crossmodal attention and multisensory integration. In C. Spence, & J. Driver (Eds.), Crossmodal space and crossmodal attention (pp. 277–320). Oxford: Oxford University Press. Stein, B. E. (2012). The new handbook of multisensory processing. Cambridge: The MIT Press. Stein, B. E., Huneycutt, W. S., & Meredith, M.A. (1988). Neurons and behavior: The same rules of multisensory integration apply. Brain Research, 448, 355–358. Stein, B. E., & Meredith, M.A. (1993). The merging of the senses. Cambridge: The MIT Press. Stratta, P., Rinaldi, O., Daneluzzo, E., & Rossi, A. (2006). Utilizzo della versione italiana del Cognitive Failures Questionnaire (CFQ) in un campione di studenti: uno studio di validazione. [The use of the Italian version of the Cognitive Failures Questionnaire (CFQ) in a sample of undergraduate students: A validation study]. Rivista di Psichiatria, 41, 260–265. Talsma, D., Senkowski, D., Soto-Faraco, S., & Woldorff, M. G. (2010). The multifaceted interplay between attention and multisensory integration. Trends in Cognitive Science, 14, 400–410. Taylor, T. L., & Donnelly, M. P. W. (2002). Inhibition of return for target discriminations: The effect of repeating discriminated and irrelevant stimulus dimensions. Perception & Psychophysics, 64, 292–317. Theeuwes, J., Van der Stigchel, S., & Olivers, C. N. (2006). Spatial working memory and inhibition of return. Psychonomic Bulletin & Review, 13, 608–613. Tipper, S. P., & Baylis, G. C. (1987). Individual differences in selective attention: The relation of priming and interference to cognitive failure. Personality and Individual Differences, 8, 667–675. Uncapher, M. R., Hutchinson, J. B., & Wagner, A.D. (2011). Dissociable effects of top-down and bottom-up attention during episodic encoding. Journal of Neuroscience, 31, 12613–12628. Van der Stigchel, S., Merten, H., Meeter, M., & Theeuwes, J. (2007). The effects of a taskirrelevant visual event on spatial working memory. Psychonomic Bulletin & Review, 14, 1066–1071. van Doorn, R. R. A., Lang, J. W. B., & Weijters, T. (2010). Self-reported cognitive failures: A core self-evaluation? Personality and Individual Differences, 49, 717–772. Wallace, J. C., Kass, S. J., & Stanny, C. J. (2002). The Cognitive Failures Questionnaire revisited: Correlates and dimensions. The Journal of General Psychology, 129, 238–256. Wheeler, M. E., Petersen, S. E., & Buckner, R. L. (2000). Memory's echo: Vivid remembering reactivates sensory-specific cortex. Proceedings of the National Academy of Sciences of the United States of America, 97, 11125–11129. Williams, M., Pouget, P., Boucher, L., & Woodman, G. F. (2013). Visual–spatial attention aids the maintenance of object representations in visual working memory. Memory and Cognition, 41, 698–715. Wolfe, J. M., Reinecke, A., & Brawn, P. (2006). Why don't we see changes? The role of attentional bottlenecks and limited visual memory. Visual Cognition, 14, 749–780.