Abstract

Voluntary orienting of visual attention is conventionally measured in tasks with predictive central cues followed by frequent valid targets at the cued location and by infrequent invalid targets at the uncued location. This implies that invalid targets entail both spatial reorienting of attention and breaching of the expected spatial congruency between cues and targets. Here, we used event-related functional magnetic resonance imaging (fMRI) to separate the neural correlates of the spatial and expectancy components of both endogenous orienting and stimulus-driven reorienting of attention. We found that during endogenous orienting with predictive cues, there was a significant deactivation of the right Temporal–Parietal Junction (TPJ). We also discovered that the lack of an equivalent deactivation with nonpredictive cues was matched to drop in attentional costs and preservation of attentional benefits. The right TPJ showed equivalent responses to invalid targets following predictive and nonpredictive cues. On the contrary, infrequent-unexpected invalid targets following predictive cues specifically activated the right Middle and Inferior Frontal Gyrus (MFG–IFG). Additional comparisons with spatially neutral trials demonstrated that, independently of cue predictiveness, valid targets activate the left TPJ, whereas invalid targets activate both the left and right TPJs. These findings show that the selective right TPJ activation that is found in the comparison between invalid and valid trials results from the reciprocal canceling of the different activations that in the left TPJ are related to the processing of valid and invalid targets. We propose that left and right TPJs provide “matching and mismatching to attentional template” signals. These signals enable reorienting of attention and play a crucial role in the updating of the statistical contingency between cues and targets.

Introduction

Deployment of attention toward upcoming spatial and temporal events is improved when these are associated with foregoing attentional cues (Posner 1980; Nobre et al. 2007). Violation of an expected cue-event association elicits an “orienting response” (Sokolov 1963) that habituates with repetition and learning of a new association (Knight and Scabini 1998; Yamaguchi et al., 2004). In natural environments, the strength of a cue-event association, that is, cue predictiveness, fluctuates between periods and contexts characterized by steady levels of high or low association to periods and contexts in which no stable probabilistic relationship is present between cues and events (Aston-Jones and Cohen 2005; Yu and Dayan 2005; Rushworth and Behrens 2008). The latter state obliges the organism to discover new statistical contingencies between cues and events.

In humans, psychophysical assessment of voluntary orienting of visual–spatial attention is conventionally achieved through the task devised by Posner (1980). This includes a high percentage (75–80%) of “valid” trials with symbolic central cues correctly indicating the location of an ensuing peripheral target. Valid trials are randomly interleaved with a low percentage of infrequent “invalid” ones (25–20%) with cues incorrectly indicating target location. It is generally believed that the strong probabilistic association between cue direction and target location is necessary to induce the use of symbolic cues. This methodology, however, has the crucial consequence that targets appearing at invalidly cued locations are infrequent. Accordingly, invalid targets do not only trigger spatial reorienting attention but they also violate the expected congruency between cue direction and target location that is induced by the higher frequency of valid trials.

A few recent functional magnetic resonance imaging (fMRI) studies started addressing the issue of the spatial and expectancy components of attentional orienting. Using short-lasting central cues that, depending on color, changed their predictive value between 60% and 90% on a trial-by-trial basis, Vossel et al. (2006) found higher activation of the right intraparietal sulcus (IPS) and of the right inferior frontal gyrus (IFG) and middle frontal gyrus (MFG) in response to infrequent invalid targets following highly predictive cues. More recently, Shulman et al. (2009) asked normal participants to shift or maintain their attention to task-relevant peripheral cues presented at target location. In different block of trials, the probability of shifting away versus maintaining attention on a cued location was randomly varied among 14%, 50%, and 86%. The study indicated that frequent-expected and infrequent-unexpected shifts of attention equally activated the right temporal–parietal junction (TPJ), thus suggesting that this area is primarily involved in spatial reorienting (Posner et al., 1984; Friedrich et al. 1998; Corbetta and Shulman, 2002) rather than in breach of expectations. On the other hand, the right IFG, anterior insula, and the caudate nuclei were sensitive to changes of expectancy, showing increased activation to infrequent-unexpected shifts of attention (Shulman et al. 2009).

These 2 studies shed some light on the spatial and expectancy components of attentional reorienting in humans; however, they reported different results as to the sensitivity of the right inferior parietal lobe to expectancy. Most importantly, in these studies, the assessment of the influence of cue predictiveness was restricted to brain activations related to orienting driven by task-relevant peripheral stimuli.

Here, we wished to gain new evidence on the neural correlates of the spatial and expectancy components of both endogenous and stimulus-driven orienting of attention. We achieved this by investigating the influence of cue predictiveness on cue- and target-related brain activity in a standard Posner task with central long-lasting arrow cues. We approached this issue by comparing blood oxygenation level–dependent (BOLD) activation in a group of participants performing the Posner task with 80% valid and 20% infrequent invalid trials, with a different group that performed the same task with equally frequent 50% valid and 50% invalid trials. Within this experimental design, the effect of invalid versus valid trials across groups should reveal brain areas involved in stimulus-driven reorienting of attention irrespective of breaches of expectation. The group (high vs. low predictiveness) by condition (invalid vs. valid trials) interaction should reveal areas involved in shifts of spatial attention specifically triggered by breaches of expectation. Critically, to highlight the specific influence of predictiveness on endogenous orienting, we also included cue-only trials, that is, trials in which the cue was not followed by any target, thus expanding previous knowledge. Varying cue predictiveness between different groups of participants allowed us to keep the structure of the task close to the classic Posner's task with fixed cue–target probabilistic relationship. This is at variance with previous studies that manipulated predictiveness within-participants, thus introducing additional strategic factors that are related to continuous trial-by-trial or block-by-block changes of cue–target probabilistic relationship (Vossel et al. 2006; Shulman et al. 2009).

The separate evaluation of cue- and target-related brain activity also allowed us clarifying possible relationships between a phenomenon originally described by Shulman et al. (2007) and endogenous orienting of spatial attention. These authors found that in a task requiring the detection of a digit-target presented within a rapid stream of letters, the right TPJ showed progressively increasing deactivation in the period preceding target occurrence. Put in other words, the more target occurrence approached and became probable along the timeline, the more TPJ was deactivated. Because TPJ deactivation was greater when ensuing targets were detected rather than missed, Shulman and coworkers proposed that right TPJ deactivation reflects efficient filtering of distracting information. Based on these observations, we hypothesized that within Posner's task, highly predictive symbolic cues should produce stronger filtering of uncued locations and greater deactivation of the right TPJ area during endogenous orienting before target presentation (i.e., in the cue phase of the trial). This would demonstrate an influence of predictiveness on preparatory, purely endogenous, processes of visuospatial attention control.

Finally, we wished to further elucidate the influence that variations in cue predictiveness might have on behavioral and brain responses to validly and invalidly cued targets. Although previous studies have shown that the reaction time (RT) advantage for validly as opposed to invalidly cued targets is greater for strongly predictive than poorly predictive or nonpredictive cues (Tipples 2002; Giessing et al. 2004; Langdon and Smith 2005; Vossel et al. 2006; Bartolomeo et al. 2007), it is still entirely unexplored whether predictiveness has equivalent or different effects on attentional benefits and costs. To investigate this additional issue, our design also comprised trials with spatially neutral cues. This allowed us to separate attentional benefits (valid vs. neutral trials) from attentional costs (invalid vs. neutral trial) and to study the influence of cue predictiveness on these.

Materials and Methods

Behavioral Task

We used a variation of the classical spatial cueing task introduced by Posner (1980), with central long-lasting arrow cues. Cues were chosen to emphasize endogenous orienting of attention based on active cue interpretation and minimizing reflexive components that can be induced by short-lasting central cues (Langdon and Smith 2005). Spatial arrangement, timing, and events of the different trial types used in the task are reported in Figure 1. Each trial began with the presentation of a central fixation cross (size: 0.8° × 0.8°) surrounded by 2 overlapping arrowheads (size 1.7° × 1.7°) and 2 lateral boxes (size 3.5° × 3.5°), one centered 7.3° to the left and the other 7.3° to the right of central fixation. One of the arrows pointed to the box on the left side and the other arrow pointed to the box on the right side. All stimuli were in white against a black background. This “Fixation” period lasted 800–1000 ms (uniform distribution) and was followed by a “Cue” period, lasting between 1400 and 1800 ms (uniform distribution). At the beginning of the Cue period, the color of arrows changed. On “directional” Valid, Invalid and “Directional Catch” trials, one arrow became yellow and the other blue. In each experimental group (i.e., “High Cue Predictiveness” vs. “No Cue Predictiveness,” see below), half of the participants were instructed to pay attention to the box pointed by the yellow arrow and the other half to the box pointed by the blue one. On nondirectional “Neutral” and “Neutral Catch” trials, the upper or lower half of each arrow became yellow and the other half blue. Corresponding upper or lower halves of the 2 arrows never had the same color (see Fig. 1). Experimental participants were explained that within nondirectional trials, central cues did not indicate any specific side of space (left/right, upper/lower), so that, in this case, they could not orient attention to one of the 2 boxes and just had to wait for target occurrence. At the end of the Cue period, a white asterisk (size: 0.4° × 0.4°) was presented as the target for 100 ms at the center of one of the 2 boxes, while the central cue remained colored. On target disappearance, the arrow cue went back white for the entire 1200-ms period allowed for response collection (“Response” period). Participants were required to hold their gaze on central fixation throughout the trial and to press a button with their right index finger as soon as possible in response to the target. On Valid trials, the target was presented in the box cued by the arrow with the relevant color. On Invalid trials, the target was presented in the box opposed to the one cued by the arrow with the relevant color. On Neutral trials, the target was presented with equal probability in one of the 2 boxes. In Directional Catch and Neutral Catch trials, no target followed cue presentation.

Figure 1.

Time course of events during experimental trials (see Materials and Methods section for details). On Directional trials, each one of the colored arrows cues one of the lateral boxes: Half of the participants were instructed to pay attention to the box pointed by the blue arrow and the other half to the box pointed by the yellow arrow. On nondirectional trials, mixed-color arrows did not allow orienting of attention to either boxes. No target was presented on Directional Catch and Neutral Catch trials.

Figure 1.

Time course of events during experimental trials (see Materials and Methods section for details). On Directional trials, each one of the colored arrows cues one of the lateral boxes: Half of the participants were instructed to pay attention to the box pointed by the blue arrow and the other half to the box pointed by the yellow arrow. On nondirectional trials, mixed-color arrows did not allow orienting of attention to either boxes. No target was presented on Directional Catch and Neutral Catch trials.

It is important to note that maintaining the cue-color code constant within each participant made any eventual working memory load negligible. Moreover, because the presentation of directional and nondirectional trials was randomized and unpredictable, any working memory load was equivalent across the different types of trials. This would not have been the case if the different types of trials were administered in separate blocks because within a block containing only nondirectional trials there would be no need to keep in memory the color code defining the direction of attention. Indeed, in this case, there would be a working memory confound when comparing directional versus nondirectional trials.

Experimental Design

Participants in the fMRI Study

Twenty-four right-handed healthy volunteers (mean age 27 years, range 23–30 years) gave written informed consent to participate in the fMRI study, which was approved by the Institutional Ethics Committee of the Fondazione Santa Lucia. Twelve participants were assigned to a High Cue Predictiveness condition (HighP) with frequent Valid and infrequent Invalid targets (80% Valid–20% Invalid trials). The other 12 participants were assigned to a No Cue Predictiveness condition (NoP) with equally frequent Valid and Invalid targets (50% Valid–50% Invalid targets). In the HighP, there were 256 Valid and 64 Invalid trials, whereas in the NoP, there were 160 Valid and 160 Invalid trials. In both conditions, there was an identical number of 160 Neutral, 64 Directional Catch, and 32 Neutral Catch trials. Trials were presented in 4 consecutive fMRI runs (144 trials per run, giving a total of 576 trials). The frequencies of each trial type, the cued side, and the target side were balanced within and across runs.

Participants from both groups were not informed on Cue predictiveness and simply asked to pay attention to the box indicated by the central cue in each trial. Before fMRI runs, they were trained to perform the task (and to interpret complex central cues), through the administration of a block of 72 trials in an environment simulating the scanner. Within this training session, Cue predictiveness was the same as in the fMRI runs. All of the 24 participants included in the fMRI study showed a reliable validity effect (i.e., faster reaction times, RTs, to Valid as compared with Invalid targets) in the training session. Four supplementary participants who did not show a validity effect in the training session did not proceed to testing inside the scanner and were excluded from the study. During fMRI runs, participants viewed stimuli through a mirror system and pressed a key with the right index finger to signal target detection.

Eye Tracking

Eye movements were continuously monitored during task performance using an ASL eye tracking system (Applied Science Laboratories, Bedford, MA; Model 504, sampling rate 60 Hz). This allowed measuring eye position during both cue and target-response periods. Off-line examination of eye position traces revealed that during both cue and target-response periods, losses of fixation (i.e., change in horizontal eye position exceeding 2° of visual angle for more than 100 ms) were absent in all participants. In all participants, eye position always remained within 0.5° of central fixation point, during both cue and target-response periods.

Imaging Protocol

A Siemens Allegra (Siemens Medical Systems, Erlangen, Germany), operating at 3 T and equipped for echoplanar imaging, acquired functional magnetic resonance images. Head movement was minimized by mild restraint and cushioning. Thirty-two slices of functional MR images were acquired using BOLD imaging (in-plane resolution = 3 × 3 mm2, slice thickness = 2.5 mm, interslice distance = 1.25 mm, time repetition = 2.08 s, time echo = 30 ms), covering the entirety of the cortex.

We used SPM5 (www.fil.ion.ucl.ac.uk/SPM) running in MATLAB 7.4 (The Math-Works Inc., Natick, MA) for data preprocessing and statistical analyses. We acquired 580 fMRI volumes for each participant. The first 4 volumes of each run were discarded to allow for T1 equilibration. After that, all images were corrected for head movement (realignment). Slice-acquisition delays were corrected using the middle slice as reference. All images were normalized to the standard SPM5 EPI template and they were spatially smoothed using an isotropic Gaussian Kernel of 8-mm full-width half-maximum. We used a 2-level analysis procedure for statistical inference.

In the first-level analysis, the time series of each participant were best fitted at every voxel using the onset timings of the 5 trial-event types in each fMRI run: Valid trials (Val), Invalid trials (Inv), Neutral trials (Neu), Directional Catch trials (DirCatch), and Neutral Catch trials (NeuCatch). The study of BOLD effects linked to the left versus right direction of cues or the side of targets was outside the scope of the present investigation and was not considered in the analysis. All stimulus-functions (Dirac's delta) were convolved with the SPM5 standard hemodynamic response function. In each participant, 5 linear contrasts estimated the mean effect (ME) of the 5 trial-event types across the 4 fMRI runs.

In a second-level analysis, individual images were entered in a 5 × 2 mixed design analysis of variance (ANOVA), modeling the 5 trial-event types separately for the 2 predictiveness conditions (i.e., for the 2 experimental groups: HighP and NoP). Within this ANOVA, we first isolated brain activations that, across the 2 experimental groups (HighP, NoP), resulted from the following contrasts: 1) Invalid versus Valid trials; 2) Invalid versus Neutral trials; 3) Valid versus Neutral trials; and 4) Directional Catch versus Neutral Catch trials. These brain areas were revealed by the ME resulting from each contrast. The rationale underlying individual contrasts is summarized at the beginning of each corresponding part of the results section. Statistical thresholds were set to P corrected <0.05 at cluster level, with cluster size estimated at P uncorrected <0.001 voxel level.

Next, within each contrast, we considered anatomical clusters of activity highlighted in the ME as Regions of Interest (ROIs). Within each ROI, we evaluated the possible influence of cue predictiveness (i.e., differences between the HighP and NoP experimental groups) by testing with the SPM toolbox “MarsBar” the statistical interaction between the 2 trial types included in the contrast and the 2 predictiveness conditions. P values were corrected for multiple comparisons (Bonferroni) as a function of the number of ROIs highlighted in the ME. When a cluster of activation included more than one anatomical–functional area (as, e.g., in the case of the cluster encompassing frontal eye fields [FEFs] that resulted from the Valid vs. Neutral comparison) separate ROIs were created by centering spherical ROIs (8-mm radius) on the local maxima of each anatomical–functional area.

Task Validation Outside the Scanner

The behavioral task used inside the scanner was previously validated outside the scanner in a separate sample of 28 right-handed participants. None of these participants was involved in the study inside the scanner. Fourteen participants were assigned to the HighP condition and 14 to the NoP condition. All of these participants showed a reliable validity effect (i.e., faster RTs to Valid as compared with Invalid targets) in the training session (1 block of 72 trials). Stimuli were displayed on a 17″ liquid crystal display monitor. Participants sat in front of the monitor at a viewing distance of 57.7 cm and responded to targets by pressing the space bar of the computer keyboard with the right index finger. The center of the screen and the keyboard were aligned to the head–body midsagittal plane of participants. Presentation of stimuli and recording of manual RTs were made with E-Prime software.

Results

Behavioral Results

Mean RTs and standard deviations (SDs) were calculated for each experimental participant. RTs exceeding 2SD around the group's mean were considered as outliers and not included in the analysis. This procedure resulted in the exclusion of less than 2% of responses.

Task Validation Outside the Scanner

Individual mean RTs were entered in a mixed ANOVA with Cue Predictiveness (HighP, NoP) as between-subjects factor and Cue type (Valid, Neutral, and Invalid) as within-subjects factor. RTs were not different in the 2 predictiveness conditions (F1,26 < 1). The analysis showed a significant main effect of Cue type (F2,52 = 39, P < 0.0001) with both significant Benefits (13.1 ms; planned Valid vs. Neutral comparison: P < 0.001) and Costs (11.1 ms; planned Invalid vs. Neutral comparison: P < 0.001). The Cue Predictiveness × Cue-type interaction was also significant (F2,52 = 3.1, P < 0.05). This interaction was further explored by evaluating the influence of Cue predictiveness on Benefits and Costs separately. A Cue Predictiveness (HighP, NoP) by Cue-type (Valid, Neutral) ANOVA revealed that attentional Benefits were present with both predictive and nonpredictive cues (Cue Predictivenss × Cue-type interaction: F1,26 = 0.26, P = 0.61; Benefits: HighP 15.8 ms, P = 0.001; NoP 11.8 ms, P = 0.001). By contrast, a Cue Predictiveness (HighP, NoP) by Cue-type (Neutral, Invalid,) ANOVA showed that with nonpredictive cues Costs were not statistically significant and were significantly reduced when compared with predictive cues (Cue Predictivenss × Cue-type interaction: F1,26 = 5.1, P = 0.03; Costs: HighP = 16.5 ms, P < 0.0001; NoP = 5.9 ms, P = ns).

Task Performance during FMRI

Behavioral data during fMRI scanning confirmed the findings from the validation study. Individual mean RTs were entered in a Cue-Predictiveness (HighP, NoP) by Cue-type (Valid, Neutral, Invalid) ANOVA. RTs were not different in the 2 predictiveness conditions (F1,22 = 2, P = 0.12). The ANOVA (see Fig. 2) showed a significant main effect of Cue type (F2,44 = 40, P < 0.0001), with both significant Benefits (39.7 ms; P < 0.001) and Costs (25.3 ms; P < 0.01). The Cue Predictiveness × Cue-type interaction was also significant (F2,44 = 5.1, P < 0.01; see Fig. 2a). A separate Cue Predictiveness (HighP, NoP) by Cue-type (Valid, Neutral) ANOVA revealed that Benefits were not influenced by Cue predictiveness (Cue Predictiveness by Cue-type interaction: F1,22 = 1.7, P = 0.19; Benefits: HighP = 46.3 ms, P < 0.001; NoP = 33.1 ms, P < 0.0001). By contrast, a Cue Predictiveness (HighP, NoP) by Cue-type (Neutral, Invalid) ANOVA showed that with nonpredictive cues Costs were not statistically significant and were significantly reduced when compared with predictive cues (Cue Predictiveness × Cue-type interaction: F1,26 = 6.2, P = 0.01; Costs: HighP = 41 ms, P < 0.001; NoP = 9.2 ms, P = 0.32).

Figure 2.

(A) Task performance during fMRI. Group-averaged RTs to Valid, Neutral, and Invalid targets in the High Cue-Predictiveness (HighP—80%Valid/20% Invalid trials) and No Cue-Predictiveness (NoP—50%Valid/50% Invalid trials) experimental groups. (B) Averaged RTs to Valid (V), Neutral (N), and Invalid (I) targets of the HighP group in the 4 consecutive blocks (B) of trials. (C) Averaged RTs to Valid, Neutral, and Invalid targets of the NoP group in the 4 consecutive blocks of trials. The NoP group shows selective abatement of attentional costs (RT difference between Invalid and Neutral trials). The performance of both groups was stable across consecutive blocks of trials.

Figure 2.

(A) Task performance during fMRI. Group-averaged RTs to Valid, Neutral, and Invalid targets in the High Cue-Predictiveness (HighP—80%Valid/20% Invalid trials) and No Cue-Predictiveness (NoP—50%Valid/50% Invalid trials) experimental groups. (B) Averaged RTs to Valid (V), Neutral (N), and Invalid (I) targets of the HighP group in the 4 consecutive blocks (B) of trials. (C) Averaged RTs to Valid, Neutral, and Invalid targets of the NoP group in the 4 consecutive blocks of trials. The NoP group shows selective abatement of attentional costs (RT difference between Invalid and Neutral trials). The performance of both groups was stable across consecutive blocks of trials.

In addition, we examined whether there was any systematic change of cueing effects (i.e., Benefits and Costs) across the 4 consecutive fMRI runs within the HighP and NoP conditions. Cue-type (Valid, Neutral, Invalid) × Block (1,2,3,4) ANOVAs revealed no significant interaction, showing that Benefits and Costs remained stable across the consecutive fMRI runs, with both predictive (HighP) and nonpredictive (NoP) cues (see Fig. 2b and c).

fMRI Results

Invalid versus Valid Trials

Within this contrast, the ME across the 2 experimental groups (HighP, NoP) defines brain areas activated by “spatial reorienting of attention.” Put in a more analytical way, this contrast isolates area activated by targets appearing at the unattended position after a directional spatial expectation was induced by the cue. This is the typical contrast that has been used in studies on attentional disengagement (Corbetta and Shulman 2002): Accordingly, with this contrast, we wished to first verify that the present study confirms patterns of cerebral activation that are usually associated with attentional disengagement.

The ME (see Fig. 3 and Table 1) showed activation of a right hemisphere network that was reported in several previous investigations. This network comprised the TPJ, IFG, and MFG ventrally, and the precuneus–superior parietal lobule (PCU–SPL) area, the IPS and FEF dorsally.

Table 1

Areas showing greater BOLD response to Invalid than Valid trials

Invalid > Valid
 
Regions P corrected Cluster size (voxels) x y z Z-score 
R FEF <0.001 1118 20 10 54 5.23 
R MFG   40 44 4.77 
*R MFG–IFG   42 12 30 4.08 
*R TPJ 0.007 316 60 −46 28 5.22 
R PCU–SPL <0.001 532 12 −56 48 5.05 
R IPS 0.005 345 34 −42 42 3.86 
Invalid > Valid
 
Regions P corrected Cluster size (voxels) x y z Z-score 
R FEF <0.001 1118 20 10 54 5.23 
R MFG   40 44 4.77 
*R MFG–IFG   42 12 30 4.08 
*R TPJ 0.007 316 60 −46 28 5.22 
R PCU–SPL <0.001 532 12 −56 48 5.05 
R IPS 0.005 345 34 −42 42 3.86 

Note: Coordinates reflect positions relative to the atlas of Tailairach and Tournoux (1988) Asterisks indicate areas showing different activation as a function of Cue Predictiveness. The MFG–IFG area showed a stronger response to infrequent (HighP condition) as compared with frequent (NoP condition) invalid targets. The TPJ area showed significant deactivation during the cue period in the HighP condition (see Directional Catch vs. neutral Catch comparison in the Results section).

Figure 3.

Areas showing greater BOLD response to Invalid than Valid trials (P <0.05 “corrected at cluster level”). The plots show the z-score of each local maxima (±90% of confidence interval) as a function of trial-type and cue-predictiveness conditions (HighP, NoP). V: valid trials, I: invalid trials, N: neutral trials, NC: neutral catch trials, and DC: directional catch trials. In each plot, the first 5 bars on the left side of the abscissa refer to the HighP condition (80% Valid/20% Invalid trials) and the 5 bars on the right side of the abscissa to the NoP condition (50% Valid/50% Invalid trials). The right TPJ showed a significant interaction of cue predictiveness (red bars with asterisk): It was more inhibited during the cue period in the HighP as compared with the NoP condition. The right mid-inferior frontal gyrus (MFG–IFG) showed a significant interaction of cue predictiveness (yellow bars with asterisk): It was more activated in response to infrequent Invalid targets in the HighP as compared with the NoP condition. The other local maxima did not show interactions as a function of cue predictiveness (green bars). TPJ: temporal–parietal junction, MFG: mid frontal gyrus, IFG: inferior frontal gyrus, MFG–IFG: mid-inferior frontal gyrus, FEF: frontal eye fields, PCU: precuneus, SPL: superior parietal lobule, and IPS: intraparietal sulcus.

Figure 3.

Areas showing greater BOLD response to Invalid than Valid trials (P <0.05 “corrected at cluster level”). The plots show the z-score of each local maxima (±90% of confidence interval) as a function of trial-type and cue-predictiveness conditions (HighP, NoP). V: valid trials, I: invalid trials, N: neutral trials, NC: neutral catch trials, and DC: directional catch trials. In each plot, the first 5 bars on the left side of the abscissa refer to the HighP condition (80% Valid/20% Invalid trials) and the 5 bars on the right side of the abscissa to the NoP condition (50% Valid/50% Invalid trials). The right TPJ showed a significant interaction of cue predictiveness (red bars with asterisk): It was more inhibited during the cue period in the HighP as compared with the NoP condition. The right mid-inferior frontal gyrus (MFG–IFG) showed a significant interaction of cue predictiveness (yellow bars with asterisk): It was more activated in response to infrequent Invalid targets in the HighP as compared with the NoP condition. The other local maxima did not show interactions as a function of cue predictiveness (green bars). TPJ: temporal–parietal junction, MFG: mid frontal gyrus, IFG: inferior frontal gyrus, MFG–IFG: mid-inferior frontal gyrus, FEF: frontal eye fields, PCU: precuneus, SPL: superior parietal lobule, and IPS: intraparietal sulcus.

The trial type (Invalid, Valid) by group (HighP, NoP) interaction showed that within the network highlighted by the ME, only the MFG–IFG area was significantly modulated by cue predictiveness. The MFG–IFG showed stronger responses to infrequent invalid targets following predictive cues (HighP) when compared with frequent invalid targets following nonpredictive cues (NoP; t = 2.77, “corrected” P = 0.02). By contrast, the right TPJ showed equivalent responses to invalid targets following predictive and nonpredictive cues and so did dorsal right PCU–SPL, IPS, and FEF areas.

Invalid versus Neutral Trials

This contrast defines brain areas related to the target-related component of attentional “costs.” Put in a more analytical way, this contrast isolates area activated in trials with targets appearing at the unattended position after a specific spatial expectation was induced by the cue versus trials in which equivalent targets appear after no spatial expectation was induced by the cue

The ME (see Fig. 4 and Table 2) showed bilateral activations in the PCU–SPL and FEF. Most important, in addition to a right TPJ cluster that was anatomically overlapping with that found in the Invalid versus Valid comparison, the Invalid versus Neutral comparison disclosed a significant cluster of activation in the left TPJ, suggesting left TPJ participation to attentional reorienting (Losier and Klein 2001). Another cluster was present in the right IFG.

Table 2

Areas showing greater BOLD response to Invalid than to Neutral trials

Invalid > Neutral
 
Regions P corrected Cluster size (voxels) x y z Z-score 
R MFG <0.001 5509 38 46 7.22 
R FEF   26 54 6.36 
L FEF   −24 54 6.29 
R TPJ <0.001 993 60 −46 28 6.44 
R PCU–SPL <0.001 1746 −60 54 6.36 
L PCU–SPL   −12 −70 58 5.01 
L Insula <0.001 593 −40 18 5.56 
L TPJ 0.001 495 −56 −48 34 4.99 
R IFG <0.001 867 52 16 −2 4.86 
R Insula   50 16 4.74 
R Putamen 0.036 214 20 10 4.59 
Invalid > Neutral
 
Regions P corrected Cluster size (voxels) x y z Z-score 
R MFG <0.001 5509 38 46 7.22 
R FEF   26 54 6.36 
L FEF   −24 54 6.29 
R TPJ <0.001 993 60 −46 28 6.44 
R PCU–SPL <0.001 1746 −60 54 6.36 
L PCU–SPL   −12 −70 58 5.01 
L Insula <0.001 593 −40 18 5.56 
L TPJ 0.001 495 −56 −48 34 4.99 
R IFG <0.001 867 52 16 −2 4.86 
R Insula   50 16 4.74 
R Putamen 0.036 214 20 10 4.59 
Figure 4.

Areas showing greater BOLD responses to Invalid than Neutral trials (P <0.05 corrected at cluster level). No area showed a significant interaction as a function of cue predictiveness.

Figure 4.

Areas showing greater BOLD responses to Invalid than Neutral trials (P <0.05 corrected at cluster level). No area showed a significant interaction as a function of cue predictiveness.

No significant trial type (Invalid, Neutral) by group (HighP, NoP) interaction was found within the network of areas individuated by the ME. Subcortically, there was bilateral activation of the Anterior Insula and unilateral activation of the right Putamen: None of these clusters was influenced by cue predictiveness.

Valid versus Neutral Trials

This contrast defines brain areas related to the target-related component of attentional “benefits.” Put in a more analytical way, this contrast isolates brain areas activated in trials with targets appearing at the attended position after a specific spatial expectation was induced by the cue versus trials in which equivalent targets appear after no spatial expectation was induced by the cue.

The ME (see Fig. 5 and Table 3) showed significant bilateral activation in the FEF, SEF IFG areas plus a TPJ cluster in the left hemisphere. The left TPJ cluster was anatomically overlapping with the left TPJ cluster highlighted by the Invalid versus Neutral comparison. This latter finding importantly shows that, when compared with Neutral trials, Invalid and Valid trials activate the same left TPJ area. This explains, in turn, the absence of any left TPJ activation in the direct Invalid versus Valid comparison.

Table 3

Areas showing greater BOLD response to Valid than to Neutral trials

Valid > Neutral
 
Regions P corrected Cluster size (voxels) x y z Z-score 
L FEF <0.001 3770 −34 −2 54 6.23 
L SEF   −4 58 6.14 
R FEF   36 48 5.91 
R Put <0.001 2016 22 10 5.78 
R IFG   50 18 −8 3.89 
L Put <0.001 2227 −16 5.67 
L IFG   −58 10 10 4.01 
*L TPJ 0.001  −52 −50 32 4.9 
L MFG 0.032 221 −36 24 38 4.05 
Valid > Neutral
 
Regions P corrected Cluster size (voxels) x y z Z-score 
L FEF <0.001 3770 −34 −2 54 6.23 
L SEF   −4 58 6.14 
R FEF   36 48 5.91 
R Put <0.001 2016 22 10 5.78 
R IFG   50 18 −8 3.89 
L Put <0.001 2227 −16 5.67 
L IFG   −58 10 10 4.01 
*L TPJ 0.001  −52 −50 32 4.9 
L MFG 0.032 221 −36 24 38 4.05 

Note: The left TPJ showed greater response to frequent (HighP condition) than to infrequent (NoP condition) Valid targets. The same interaction was present in the left IFG at P uncorrected at level (P = 0.01; not reported in the table).

Figure 5.

Areas showing greater BOLD responses to Valid than Neutral trials (P< 0.05 corrected at cluster level). The left TPJ was more activated by targets in the HighP as compared with the NoP condition (yellow bars with asterisk). The left IFG showed a similar interaction that was significant at P uncorrected level (P = 0.01). No equivalent interaction was found in all other areas. SEF: supplementary eye fields.

Figure 5.

Areas showing greater BOLD responses to Valid than Neutral trials (P< 0.05 corrected at cluster level). The left TPJ was more activated by targets in the HighP as compared with the NoP condition (yellow bars with asterisk). The left IFG showed a similar interaction that was significant at P uncorrected level (P = 0.01). No equivalent interaction was found in all other areas. SEF: supplementary eye fields.

The trial type (Valid, Neutral) by group (HighP, NoP) interaction showed that within the network individuated by the ME, the left TPJ (t = 3.1, corrected P = 0.01) responded more strongly to frequent valid targets following predictive cues (HighP condition) than to valid targets following nonpredictive cues (NoP condition). The same trend was found in the left IFG (t = 2.1, uncorrected P = 0.01). Subcortically, the Putamen was bilaterally activated and not modulated by cue predictiveness.

Directional versus Neutral Catch Trials

Both Directional and Neutral Catch trials did not include target presentation: This allows eliminating any target-related component of attentional orienting and isolating brain activations specifically related to endogenous orienting with central symbolic cues. The contrast between Directional and Neutral Catch trials, defines areas activated by spatially directional cues, that is, cues that bias endogenous orienting toward one of the 2 possible target locations versus areas activated by spatially nondirectional cues, that is, cues that do not bias attention toward a specific target location.

The ME of this comparison showed a bilateral cluster of activations in the SEF and a cluster in the left FEF (see Fig. 6 and Table 4). Two other clusters were individuated at P uncorrected cluster level, in the right FEF (P = 0.01; cluster size = 138 voxels) and left IPS (P = 0.02; cluster size 113 voxels). The trial type (Directional Catch, Neutral Catch) by group (HighP, NoP) interaction revealed that none of the clusters highlighted by the ME was influenced by cue predictiveness.

Table 4

Areas showing greater BOLD response to Directional Catch than to Neutral Catch trials

Directional Catch > Neutral catch
 
Regions P corrected Cluster size (voxels) x y z Z-score 
L FEF <0.001 916 −22 -50 4.28 
L SEF   −6 68 4.19 
R SEF   10 70 3.59 
Directional Catch > Neutral catch
 
Regions P corrected Cluster size (voxels) x y z Z-score 
L FEF <0.001 916 −22 -50 4.28 
L SEF   −6 68 4.19 
R SEF   10 70 3.59 

Note: Two additional clusters, not reported in the table, were found in the left IPS and the right FEF at P uncorrected at cluster level (P = 0.02 and P = 0.01, respectively).

Figure 6.

Areas showing greater BOLD responses during endogenous attentional orienting, as resulting from the comparison between Directional Catch and Neutral Catch trials (P <0.05 corrected at cluster level). Clusters in the left IPS and the right FEF were significant at P uncorrected at cluster level (P = 0.02 and P = 0.01, respectively). None of these areas showed an interaction with cue predictiveness.

Figure 6.

Areas showing greater BOLD responses during endogenous attentional orienting, as resulting from the comparison between Directional Catch and Neutral Catch trials (P <0.05 corrected at cluster level). Clusters in the left IPS and the right FEF were significant at P uncorrected at cluster level (P = 0.02 and P = 0.01, respectively). None of these areas showed an interaction with cue predictiveness.

We further investigated the role of predictiveness in the cueing phase considering also possible patterns of “deactivation.” In fact, Shulman et al. (2007) found that the right TPJ area that in their study showed significant responses to targets presented in 75% of trials also showed progressively increasing deactivation in the period preceding target occurrence. Based on this evidence, we tested whether the right TPJ area that in our investigation showed equivalent BOLD response to infrequent versus frequent invalid targets (i.e., Invalid vs. Valid comparison) was differentially deactivated during the directional cue period as a function of cue predictiveness. We considered all significant ROIs resulting from the Invalid versus Valid comparison and tested the between group interactions in the Directional Cue versus Neutral Cue contrast (see Materials and Methods). A significant interaction was found only in the right TPJ, which was significantly more deactivated during orienting with highly predictive directional cues (HighP condition) when compared with nonpredictive ones (NoP condition; t = 2.89, corrected P = 0.01). See Figure 3 and Table 1.

At variance with recent findings reported by Shulman et al. (2009), we did not find modulation by cue predictiveness on cue-related activations in dorsal attentional areas (SPL, IPS, and FEF). Because in our study participants were not informed of cue predictiveness, it could be hypothesized that predictiveness was progressively discovered during the consecutive block of trials and that cue-related brain activity changed accordingly, with activations in the late blocks resembling activations documented by Shulman et al. (2009) in informed participants.

We explored this possibility comparing cue-related activity in the first half (fMRI runs 1 and 2) versus second half (fMRI runs 3 and 4) of the experiment. For each subject, the relevant contrast “Directional versus Neutral Catch trials” was computed separately for the first and second halves of the experiment and submitted to a 2 × 2 ANOVA with the factors “Part of the experiment” (first half, second half) and Group (HighP, NoP). Using the areas highlighted by the ME in the Directional versus Neutral Catch comparison (cf. Fig. 6), we then tested for the ME of the Part of the experiment and for the critical Part of the experiment (first, second) by Group (HighP, NoP) interaction in the new analysis. These additional tests did not reveal significant effects, other than the main effect of Directional versus Neutral Catch, in any of the 5 regions activated by directional cues (i.e., left FEF, left and right SEFs, plus the right FEF and the left IPS).

This finding emphasizes that cue-related activity did not change across the consecutive block of trials in both cue-predictiveness conditions. We conclude that though not explicitly informed on cue predictiveness, participants were able to implicitly set a stable attentional strategy within the training block of 72 trials performed before entering the scanner. This attentional strategy was maintained throughout the 4 runs of trials within the scanner, as also indicated by the absence of significant differences in the psychophysical performance across the runs in both cue-predictiveness conditions.

Discussion

The study of brain activations that in each trial of a Posner task are specifically related to the endogenous-cueing phase or to the ensuing detection of targets is of particular relevance because endogenous orienting of attention engages a right hemispheric dorsal network (comprising the SPL, the IPS, and the FEFs), which is anatomically and functionally separated from a more ventral network (comprising the TPJ and IFG areas) that is engaged by reorienting to unattended/unexpected targets (Friedrich et al. 1998; Gitelman et al. 1999; Husain and Rorden 2003; Corbetta et al. 2008; Singh-Curry and Husain 2009). Here, by separating cue- and target-related BOLD activations, we isolated the neural correlates of the spatial and expectancy components of endogenous and stimulus-driven orienting of attention. The inclusion of trials with neutral cues, which have been rarely considered in fMRI studies of spatial attention, allowed us to explore the neural correlates of the endogenous and stimulus-driven components of attentional benefits and costs and the effects of expectancy on these. On the whole, our observations disclose new evidence on the hemispheric lateralization of the endogenous control of attention and of brain responses to valid and invalid targets. This allows drawing a number of conclusions and advance new hypotheses on mechanisms of attentional control.

TPJ and MFG: The Spatial and Expectancy Components of Stimulus-Driven Reorienting

The present study shows that unexpected-infrequent and expected-frequent invalid targets that trigger identical spatial operations of attentional reorienting, evoke equivalent BOLD responses in the TPJ, SPL, and FEF areas in the right hemisphere. By contrast, unexpected-infrequent invalid targets determine a specific enhancement of the BOLD response in the right MFG–IFG area when compared with expected-frequent ones. These 2 main findings delineate, respectively, the neural bases of the spatial and expectancy components of attentional reorienting in humans. The patterns of target-related activity that we have found in the right TPJ and MFG–IFG areas are consistent with those that Shulman et al. (2009) have observed studying brain responses to infrequent and frequent shifts of attention toward peripheral cues.

The functional significance of the TPJ and IFG–MFG responses to invalid targets can be better understood in the context of the electrophysiological evidence, in particular the evoked potential known as the P3. Unexpected-infrequent targets typically elicit a positive electrophysiological response peaking at around 300-ms poststimulus (P3; Knight and Scabini 1998). The early component of this response (P3a) is generated in the prefrontal areas and is related to the orienting reaction toward “novel” stimuli (Daffner et al. 2000, 2003). Enhanced activation of the IFG–MFG area in response to infrequent invalid targets following highly predictive cues can therefore provide a novelty-alerting signal that helps interruption of endogenous orienting and reorienting of attention in dorsal and ventral attentional networks (Corbetta et al. 2008). We argue that this can be particularly useful when statistical prevalence of valid trials over invalid ones induces strong focusing of endogenous attention on cued locations.

The TPJ generates a late component of the P3 response, the P3b (Knight and Scabini 1998; Daffner et al. 2000; 2003). This component reflects the categorization of the stimulus as a function of its matching to an internal model, which helps, in turn, updating of model representation in working memory (Donchin and Coles 1988). In the light of this evidence, the response of the right TPJ to invalid targets can be interpreted as being generated by the mismatch between cued and actual target location. Here, we have found that this mismatch response is not modulated by cue predictiveness, because infrequent and frequent invalid targets determined equivalent BOLD responses in the right TPJ.

The Contributions of the Left and Right TPJs in Stimulus-Driven Reorienting: Matching and Mismatching to an Attentional Template

New hints on the functional role of the TPJ in reorienting of attention are offered by the results of BOLD comparisons of Valid and Invalid trials with nondirectional Neutral trials. The Valid versus Neutral comparison revealed target-related activity in the left TPJ: Since all other target-related factors are equal between these 2 types of trials, this activation seems specifically linked to the match between expected and actual target position characterizing Valid trials. In a similar vein, the Invalid versus Neutral comparison showed bilateral activation of the TPJ: this activation can be traced back to the mismatch between expected and actual target position characterizing Invalid trials. It is worth emphasizing that TPJ activation was absent in the Directional versus Neutral cues comparison and that TPJ deactivation, rather than activation, was present during orienting with predictive cues. Both of these findings further suggest that TPJ activations resulting from the Valid versus Neutral and the Invalid versus Neutral comparisons are target related rather than cue related. Evoked response potentials investigations indicated that the role of the TPJ is to coordinate a “template Matching” (tM) for task-relevant stimuli (Donchin and Coles 1988; Knight and Scabini 1998; Daffner et al. 2000, 2003). It is easy to argue that within the Posner task, comparing actual with cued target position does not only help shifts of attention but also allows for dynamic updating of internal models of cue–target contingency. This latter function is relevant for keeping or switching the attentional-goal set determined by probabilistic cue–target contingency (Aston-Jones and Cohen 2005; Corbetta et al. 2008).

The present study provides new evidence suggesting the anatomical–functional segregation of a tM system in the left TPJ, as indicated by the results of the Valid versus Neutral comparison, and a “template-MisMatching” (tMisM) system in the TPJ areas of both hemispheres as indicated by the results of the Invalid versus Neutral comparison. Notwithstanding a number of exceptions (Downar et al. 2001; Serences et al. 2005; Indovina and Macaluso 2007; Macaluso and Patria 2007; Natale et al. 2008), in many previous studies, the response of the left TPJ to invalid targets went undetected. Results from the Valid versus Neutral comparison, shows that this was due to the fact that the left TPJ also contains neuronal populations responding to valid targets. This has the obvious consequence that left TPJ activations related to cue–target match on Valid trials and cue–target mismatch on Invalid trials, reciprocally cancel out in the direct Invalid versus Valid comparison.

The Valid versus Neutral comparisons revealed enhancement of the left TPJ and the left IFG response to frequent valid targets that followed predictive cues. Interestingly, this target-related sensitivity of the left TPJ–IFG area might account for residual abilities of right brain–damaged patients with left spatial neglect (Doricchi et al. 2008) to implicitly exploit statistical contingencies governing the spatial distribution of targets (Bartolomeo et al. 2001; Geng and Behrmann 2002).

TPJ, Reorienting of Attention with Endogenous or Exogenous Cues and Contingent Attentional Capture

It was emphasized that the latency of the P3 is longer than latencies of responses to visual stimuli in dorsal parietal areas and that, therefore, reorienting to invalid targets might not be initiated by the TPJ (Corbetta et al. 2008). We found of particular interest that reorienting of attention after invalid central cues does activate the right TPJ, whereas reorienting after invalid exogenous peripheral cues does not activate the same area (Kincade et al. 2005; Natale et al. 2008). Based on our interpretation of the TPJ role, this finding suggests that a basic difference between exogenous and endogenous orienting is that only in the latter case a template of the expected attentional event is prepared during the cue period and then compared, through a tM/tMisM process, with the actual attentional event in the target period.

The localization in the right and left TPJ of tM and tMisM neuronal populations can also suggest a new interpretation of contingent attentional capture (Serences et al. 2005). When irrelevant stimuli share a feature with relevant attentional targets, they can determine contingent capturing of attention (Serences et al. 2005). Contingent attentional shifts activate the right and, to a lesser degree, the left TPJ (Serences et al. 2005). This was taken as evidence that the TPJ promotes shifts of attention toward all stimuli possessing features defined in attentional settings. Here, we wish to refine this interpretation by proposing that the TPJ activation by contingent attentional capture originates from the mismatch produced by the occurrence of attentionally relevant features at unattended locations. As an example, when somebody is looking for a red letter within a stream of letters presented at central fixation (Serences et al. 2005), the TPJ activation produced by a red letter occurring at one side of fixation depends on the occurrence of a target-defining feature at a location mismatching the one specifically attended by the participant.

TPJ, Cue Predictiveness and Endogenous Orienting

By including a condition where no target stimulus followed the cue, we were able to isolate the specific influence of predictiveness on cue-related TPJ activity. We have found that during the cue period, the right TPJ is significantly deactivated when directional cues are highly predictive and invalid targets infrequent. Shulman et al. (2007) have proposed that signals for task relevance deactivate the right TPJ, filtering out distracting stimuli. Our findings suggest that cue predictiveness sets the level to which the right TPJ is prepared to code a mismatch between expected and actual target locations. The higher the cue predictiveness, the stronger the focusing of attention on cued location, the stronger the filtering out of uncued location, the stronger the TPJ deactivation, and the weaker the preparation of attentional networks to code a mismatch between expected and actual target location. The level of filtering activity in the TPJ might be regulated by the history of cue effectiveness during task performance. When cue predictiveness is high (HighP), the use of cues is rewarded and invalid targets trigger the reorienting response in the TPJ infrequently: both of these factors might underlie TPJ deactivation during the cue period. On the other hand, when cue predictiveness is low or absent (NoP), the use of cues is not rewarded and invalid targets trigger the reorienting response frequently: both of these factors might favor maintenance of steady activation in the right TPJ and facilitation of attentional reorienting. Setting of TPJ filtering activity might depend on signals from the Anterior Cingulate) and Orbital Frontal cortex, which assess the reward provided by the use of cues (Aston-Jones and Cohen 2005; Rushworth and Behrens 2008). Both of these structures send efferents to the Locus Coeruleus in the brainstem which, in turn, can exert its noradrenergic modulation through efferents innervating the TPJ area (Aston-Jones and Cohen 2005; Corbetta et al. 2008).

Cue Predictiveness Modulates Attentional Costs Selectively

Lack of TPJ deactivation during endogenous orienting with nonpredictive cues was matched, on target occurrence, to selective abatement of attentional costs (RT difference between Invalid and Neutral trials) and upholding of attentional benefits (RT difference between Neutral and Valid trials). Importantly, this finding points to independent modulation by cue predictiveness of behavioral responses to invalid targets. It is worth noting that this type of modulation allows speeding up of reorienting in environmental contexts characterized by weak probabilistic relationship between cues and targets and, at the same time, preservation of attentional benefits provided by valid cues within the very same contexts.

Cue Predictiveness and Orienting of Attention in Dorsal Frontal–Parietal Areas

In the present study, we found that expectancy did not modulate BOLD responses in both dorsal attentional areas activated by reorienting to invalid targets, that is, right PCU–SPLand FEF, and in dorsal areas activated during endogenous orienting with central directional symbolic cues, that is, IPS, SEF, and FEF. These findings do not seem in line with recent observations by Shulman et al. (2009). These authors found that breaches of expectation enhanced BOLD activity in both dorsal areas showing transient responses to peripheral cues inducing shifts of attention (the left and right PCU–SPL , right FEF) and in areas showing sustained responses during maintenance of attention on the same cues (the left and right IPS). Relevant differences between the experimental design adopted in our study and that used by Shulman et al. (2009) might account for the different results. In the present investigation, cue predictiveness was varied between different groups of participants, whereas it remained constant, as in the classic Posner's task, across the consecutive blocks of trials performed by each participant. On the contrary, in the study by Shulman et al. (2009), predictiveness of peripheral cues was randomly alternated among 14%, 50%, and 86% across the 20 blocks of trials administered to each participant. Further, in the present study, participants were not informed on cue predictiveness, whereas in the study by Shulman and coworkers, participants were informed as to the predictiveness assigned to each block of trials. Different cognitive processes might therefore have been tapped in the 2 studies. For example, in the study by Shulman et al. (2009), knowing what probability of shifting versus maintaining attention was to be expected in the block of trials to come might have induced specific preparatory decision processes as to the prevalent selection of “shift” versus “stay” attentional responses. This might have influenced the activity of dorsal attentional areas that are known to modify their level of activity toward the most probable stimulus–response association within the set of associations that can be generated in a sensory-motor task (Platt and Glimcher 1999; Sugrue et al. 2004; Heekeren et al. 2008; Tosoni et al. 2008).

We also considered the possibility that because in our study participants were not informed on cue predictiveness, learning cue predictiveness during the consecutive blocks of trials might have increased variance in BOLD responses both between and within individual participants. This, in turn, might have obscured relevant differences between early and late blocks of trials, with BOLD activity in late blocks, that is, when participants had supposedly well-recognized cue predictiveness, resembling more the results reported by Shulman et al. (2009) on informed participants. This hypothesis, however, was not supported by the absences of significant changes in psychophysical performance across consecutive blocks of trials, both inside and outside the scanner, and by the absence of significant changes in BOLD responses recorded in the first (runs 1 and 2) and the second (runs 3 and 4) halves of the experiment in dorsal areas engaged by endogenous orienting with central directional cues.

Conclusions

In summary, our findings dissociate the neural correlates of the spatial and expectancy components of endogenous and stimulus-driven orienting of attention in humans. The main findings were as follows: First, during endogenous orienting with symbolic cues, the right TPJ regulates focusing of attention on cued location as a function of cue predictiveness: The higher the predictiveness, the greater right TPJ deactivation and the filtering out of the uncued location. We also discovered that reduced filtering out of the uncued location with nonpredictive cues produces a drop in attentional costs without affecting attentional benefits. Second, the activity of the right IFG–MFG area is enhanced by breaches of expectation produced by infrequent invalid targets following highly predictive cues. Third, the left and the right TPJ code mismatches between cued and actual target location independently of cue predictiveness. Fourth, the left TPJ also code matches between expected and actual target locations. We argue that the match–mismatch function of the TPJ areas helps reorienting of attention and has a crucial role in the monitoring of cue–target association whose variations promote the keeping or the switching of the attentional task set.

Funding

Ministero della Universita’e della Ricerca Scientifica e Tecnologica and Fondazione Santa Lucia Istituto di Ricovero e Cura a Carattere Scientifico (to F.D.). The Italian Ministry of Health to Neuroimaging Laboratory of the Fondazione Santa Lucia.

We wish to thank C. Rossi-Arnaud, E. Natale, A. Chica, M. Thiebaut de Schotten, P. Bartolomeo, S. Fairhall, and 2 anonymous referees for suggestions and discussions. Conflict of Interest: None declared.

References

Aston-Jones
G
Cohen
JD
An integrative theory of locus coeruleus–norepinephrine function: adaptive gain and optimal performance
Annu Rev Neurosci
 , 
2005
, vol. 
28
 (pg. 
403
-
450
)
Bartolomeo
P
Decaix
C
Siéroff
E
The phenomenology of endogenous orienting
Conscious Cogn
 , 
2007
, vol. 
16
 (pg. 
144
-
161
)
Bartolomeo
P
Siéroff
E
Decaix
C
Chokron
S
Modulating the attentional bias in unilateral neglect: the effects of the strategic set
Exp Brain Res
 , 
2001
, vol. 
137
 (pg. 
432
-
444
)
Corbetta
M
Patel
G
Shulman
GL
The reorienting system of the human brain: from environment to theory of mind
Neuron
 , 
2008
, vol. 
58
 (pg. 
306
-
324
)
Corbetta
M
Shulman
GL
Control of goal-directed and stimulus-driven attention in the brain
Nat Rev Neurosci
 , 
2002
, vol. 
3
 (pg. 
201
-
215
)
Daffner
KR
Mesulam
MM
Scinto
LF
Acar
D
Calvo
V
Faust
R
Chabrerie
A
Kennedy
B
Holcomb
P
The central role of the prefrontal cortex in directing attention to novel events
Brain
 , 
2000
, vol. 
123
 (pg. 
927
-
939
)
Daffner
KR
Scinto
LF
Weitzman
AM
Faust
R
Rentz
DM
Budson
AE
Holcomb
PJ
Frontal and parietal components of a cerebral network mediating voluntary attention to novel events
J Cogn Neurosci
 , 
2003
, vol. 
15
 (pg. 
294
-
313
)
Donchin
E
Coles
MGH
Is the P300 component a manifestation of context updating?
Behav Brain Sci
 , 
1988
, vol. 
11
 (pg. 
357
-
374
)
Doricchi
F
Thiebaut de Schotten
M
Tomaiuolo
F
Bartolomeo
P
White matter (dis)connections and gray matter (dys)functions in visual neglect: gaining insights into the brain networks of spatial awareness
Cortex
 , 
2008
, vol. 
44
 (pg. 
983
-
995
)
Downar
J
Crawley
AP
Mikulis
DJ
Davis
KD
The effect of task elevance on the cortical response to changes in visual and auditory stimuli: an event-related fMRI study
NeuroImage
 , 
2001
, vol. 
14
 (pg. 
1256
-
1267
)
Friedrich
FJ
Egly
R
Rafal
RD
Beck
D
Spatial attention deficits in humans: a comparison of superior parietal and temporal-parietal junction lesions
Neuropsychology
 , 
1998
, vol. 
12
 (pg. 
193
-
207
)
Geng
JJ
Behrmann
M
Probability cuing of target location facilitates visual search implicitly in normal participants and patients with hemispatial neglect
Psychol Sci
 , 
2002
, vol. 
13
 (pg. 
520
-
525
)
Giessing
C
Thiel
CM
Stephan
KE
Rfsler
F
Fink
GR
Visuospatial attention: how to measure effects of infrequent, unattended events in a blocked stimulus design
Neuroimage
 , 
2004
, vol. 
23
 (pg. 
1370
-
1381
)
Gitelman
DR
Nobre
AC
Parrish
TB
LaBar
KS
Kim
YH
Meyer
JR
Mesulam
MM
A large scale distributed network for covert spatial attention. Further anatomical delineation based on stringent behavioural and cognitive controls
Brain
 , 
1999
, vol. 
122
 (pg. 
1093
-
1106
)
Heekeren
HR
Marrett
S
Ungerleider
LG
The neural systems that mediate human perceptual decision making
Nat Rev Neurosci
 , 
2008
, vol. 
9
 (pg. 
467
-
479
)
Husain
M
Rorden
C
Non-spatially lateralized mechanisms in hemispatial neglect
Nat Rev Neurosci
 , 
2003
, vol. 
4
 (pg. 
26
-
36
)
Indovina
I
Macaluso
E
Dissociation of stimulus relevance and saliency factors during shifts of visuospatial attention
Cereb Cortex
 , 
2007
, vol. 
17
 (pg. 
1701
-
1711
)
Kincade
JM
Abrams
RA
Astafiev
SV
Shulman
GL
Corbetta
M
An event-related functional magnetic resonance imaging study of voluntary and stimulus-driven orienting of attention
J Neurosci
 , 
2005
, vol. 
25
 (pg. 
4593
-
4604
)
Knight
RT
Scabini
D
Anatomic bases of event-related potentials and their relationship to novelty detection in humans
J Clin Neurophysiol
 , 
1998
, vol. 
15
 (pg. 
3
-
13
)
Langdon
R
Smith
P
Spatial cueing by social versus nonsocial directional signals
Visual Cog
 , 
2005
, vol. 
12
 (pg. 
1497
-
1527
)
Losier
BJW
Klein
RM
A review of the evidence for a disengage deficit following parietal lobe damage
Neurosci Biobehav Rev
 , 
2001
, vol. 
25
 (pg. 
1
-
13
)
Macaluso
E
Patria
F
Spatial re-orienting of visual attention along the horizontal or the vertical axis
Exp Brain Res
 , 
2007
, vol. 
180
 (pg. 
23
-
34
)
Natale
E
Marzi
CA
Macaluso
E
FMRI correlates of visuo-spatial reorienting investigated with an attention shifting double-cue paradigm
Hum Brain Mapp
 , 
2008
 
[Nov 25: Epub ahead of print]
Nobre
AC
Correa
A
Coull
JT
The hazards of time
Curr Opin Neurobiol
 , 
2007
, vol. 
17
 (pg. 
465
-
470
)
Platt
ML
Glimcher
PW
Neural correlates of decision variables in parietal cortex
Nature
 , 
1999
, vol. 
400
 (pg. 
233
-
238
)
Posner
MI
Orienting of attention
Q J Exp Psychol
 , 
1980
, vol. 
32
 (pg. 
3
-
25
)
Posner
MI
Walker
JA
Friedrich
FJ
Rafal
RD
Effects of parietal injury on covert orienting of attention
J Neurosci
 , 
1984
, vol. 
4
 (pg. 
1863
-
1874
)
Rushworth
MFS
Behrens
TEJ
Choice, uncertainty and value in prefrontal and cingulate cortex
Nat Rev Neurosci
 , 
2008
, vol. 
11
 (pg. 
390
-
397
)
Serences
JT
Shomstein
S
Leber
AB
Golay
X
Egeth
HE
Yantis
S
Coordination of voluntary and stimulus-driven attentional control in human cortex
Psychol Sci
 , 
2005
, vol. 
16
 (pg. 
114
-
122
)
Shulman
GL
Astafiev
SV
Franke
D
Pope
DL
Snyder
AZ
McAvoy
MP
Corbetta
M
Interaction of stimulus-driven reorienting and expectation in ventral and dorsal frontoparietal and basal ganglia-cortical networks
J Neurosci
 , 
2009
, vol. 
29
 (pg. 
4392
-
4407
)
Shulman
GL
Astafiev
SV
McAvoy
MP
d'Avossa
G
Corbetta
M
Right TPJ deactivation during visual search: functional significance and support for a filter hypothesis
Cereb Cortex
 , 
2007
, vol. 
17
 (pg. 
2625
-
2633
)
Singh-Curry
V
Husain
M
The functional role of the inferior parietal lobe in the dorsal and ventral stream dichotomy
Neuropsychologia
 , 
2009
, vol. 
47
 (pg. 
1434
-
1448
)
Sokolov
EN
Higher nervous functions: the orienting reflex
Annu Rev Physiol
 , 
1963
, vol. 
25
 (pg. 
545
-
580
)
Sugrue
LP
Corrado
GS
Newsome
WT
Matching behavior and the representation of value in the parietal cortex
Science
 , 
2004
, vol. 
304
 (pg. 
1782
-
1787
)
Talairach
J
Tournoux
P
Co-planar stereotaxic atlas of the human brain
 , 
1988
New York
Thieme Medical Publishers, Inc
Tipples
J
Eye gaze is not unique: automatic orienting in response to uninformative arrows
Psychonom Bull Rev
 , 
2002
, vol. 
9
 (pg. 
314
-
318
)
Tosoni
A
Galati
G
Romani
GL
Corbetta
M
Sensory-motor mechanisms in human parietal cortex underlie arbitrary visual decisions
Nat Neurosci
 , 
2008
, vol. 
11
 (pg. 
1446
-
1453
)
Vossel
S
Thiel
CM
Fink
GR
Cue validity modulates the neural correlates of covert endogenous orienting of attention in parietal and frontal cortex
Neuroimage
 , 
2006
, vol. 
32
 (pg. 
1257
-
1264
)
Yamaguchi
S
Hale
LA
D'Esposito
M
Knight
R
Rapid prefrontal–hyppocampal habituation to novel events
J Neurosci
 , 
2004
, vol. 
24
 (pg. 
5356
-
5363
)
Yu
AJ
Dayan
P
Uncertainty, neuromodulation and attention
Neuron
 , 
2005
, vol. 
46
 (pg. 
681
-
692
)