Psychological Research DOI 10.1007/s00426-014-0635-8

ORIGINAL ARTICLE

Timing the events of directional cueing Giovanna Girardi • Gabriella Antonucci Daniele Nico

•

Received: 8 September 2014 / Accepted: 25 November 2014  Springer-Verlag Berlin Heidelberg 2014

Abstract To explore the role of temporal context on voluntary orienting of attention, we submitted healthy participants to a spatial cueing task in which cue-target stimulus onset asynchronies (SOAs) were organized according to two-dimensional parameters: range and central value. Three ranges of SOAs organized around two central SOA values were presented to six groups of participants. Results showed a complex pattern of responses in terms of spatial validity (faster responses to correctly cued target) and preparatory effect (faster responses to longer SOAs). Responses to validly and neutrally cued targets were affected by the increase in SOA duration if the difference between longer and shorter SOA was large. On the contrary, responses to invalidly cued targets did not vary according to SOA manipulations. The observed pattern of cueing effects does not fit in the typical description of spatial attention working as a mandatory disengaging– shifting–engaging routine. In contrast, results rather suggest a mechanism based on the interaction between context sensitive top-down processes and bottom-up attentional processes.

G. Girardi
G. Antonucci
D. Nico Department of Psychology, Sapienza University of Rome, Rome, Italy G. Girardi (&) Institut Jean Nicod, CNRS, Ecole Normale Supe´rieure, 29, rue d’Ulm, 75005 Paris, France e-mail: [email protected] G. Antonucci Centro Ricerche Neuropsicologia, Santa Lucia Foundation IRCCS, Rome, Italy

Introduction Spatial attention is often described as a focal point that is oriented to different locations in space to inspect the visual world. Implicit in this concept is the idea that a single region of the visual field is selected, which receives priority in information processing. This region has been referred to as the observer’s attentional window (Nakayama, 1990), a concept similar to the spotlight (Posner & Petersen, 1990) or zoom-lens (Eriksen & St. James, 1986) metaphors. Past (Posner, Cohen, & Rafal, 1982) and present theories of spatial attention (Posner, Cohen, & Rafal, 2008) maintain that prior to shifting, attention must be disengaged to be reengaged to the intended destination at the end of the shift. This precise sequence of operations is thought to relate to a set of brain regions each involved in a single step of this functional routine (Posner et al., 1982; Raz & Buhle, 2006). In experimental settings, spatial orienting of attention is typically initiated by providing directional cues (i.e., an arrow) that point towards the probable location of the upcoming targets. Presentation of a target following a valid cue significantly reduces reaction times (RTs). Conversely, a cue suggesting an invalid location increases RTs because the current focus of attention must be reoriented to the target’s effective position (Posner, Petersen, Fox, & Raichle, 1988; Posner, Snyder, & Davidson, 1980). The operational routine of disengaging/shifting/engaging attention is considered a mandatory process that is involved whenever a directional spatial cue is presented. Shifting of attention is not modulated only by the explicit directional information provided by the cue. For instance, involuntary spatial orienting of attention is triggered by the semantic-physical properties of the arrow’s head (Hommel, Pratt, Colzato, & Godijn, 2001; Ranzini,

123

Psychological Research

Dehaene, Piazza, & Hubbard, 2009; Tipples, 2002). Moreover, independently from its meaning and physical properties, the mere onset of a cue recruits attention as it implicitly conveys the information that the target will soon appear. This temporal component has been linked to alerting, arousal, and foreperiod effects that occur whenever a warning signal is presented prior to a target (Fan, McCandliss, Sommer, Raz, & Posner, 2002; Gabay & Henik, 2010; Los & Schut, 2008; Mu¨ller-Gethmann, Ulrich, & Rinkenauer, 2003; Posner & Boies, 1971; Posner & Petersen, 1990). The interval of time between the cue and the target (stimulus onset asynchrony, SOA) can also modulate attention when several SOAs are included in a block of trials. Numerous experiments investigating response readiness to targets preceded by a warning stimulus (i.e., the foreperiod, FP) have widely demonstrated that detection is improved if the target is actually shown at the moment when participants estimate its appearance as most likely (e.g., Alegria & Bertelson, 1970; Drazin, 1961; Elithorn & Lawrence, 1955; Karlin, 1959; Klemmer, 1956; Na¨a¨ta¨nen, 1970, 1972; Sanders, 1966; Thomas, 1967). Moreover, when FPs of different lengths are presented with equal frequency within a single block of trials, RTs are typically found to decrease as a function of FP duration (i.e., the socalled FP effect, see Niemi & Na¨a¨ta¨nen, 1981, for a review). Researchers related this finding to the fact that the conditional probability of an event (in this case, target onset) increases along with the progressive elapsing of time. The growing probability would then boost recruitment of preparatory or attentional resources and speed up the reaction time. This would explain the FP effect (Cui, Stetson, Montague, & Eagleman, 2009; Niemi & Na¨a¨ta¨nen, 1981; Nobre, Correa, & Coull, 2007). More recently, an effect sensitive to the context of the intervals used in the other trials has been shown to influence response to spatially cued targets. In particular, the facilitation provided by an informative spatial cue was found to additionally depend on SOAs’ relative frequency, i.e., on temporal contingencies across (rather than within) trials (Girardi, Antonucci, & Nico, 2013). Indeed, a valid spatial cue improved participants’ performance only in trials in which target onset was highly predictable because of the higher occurrence of the corresponding SOA. Conversely, cueing proved ineffective when cue and target were associated to a less frequent temporal asynchrony. These results demonstrated that voluntary control of attention may be switched on and off within a given trial, according to probability of target presentation in the entire trials’ set. The notion that spatial focus can be controlled and dynamically modulated depending on preparatory strategies is inconsistent with two basic assumptions of the

123

standard attentional metaphors. The first one is that trials are independent from one another. This implies that each trial in a block provides independent measures of attention allocation. The second assumption is that spatial orienting and alerting are independent processes (e.g., Posner & Petersen, 1990). At present, growing evidence from independent research lines challenges both assumptions. With respect to trials’ independence, it has been shown that when people are presented with targets that differ in their probability of occurrence, cueing effects are found only for the more probable target (e.g., Klein & Hansen, 1990). Similarly, Lambert, Norris, Naikar, and Aitken (1999) showed that participants could benefit from predictive information about the location of the target when such information was learned as a cue-target relationship implicitly conveyed through presentation of many trials. The possibility that voluntary spatial orienting in one trial may be sensitive to orienting processes in the previous trial has been also investigated. Jongen and Smulders (2007) found that in a directional cueing paradigm, the spatial validity effect in a specific trial was larger when the previous trial included a valid rather than an invalid cue. Furthermore, contrary to the view that only one kind of process guides attentional orienting on a trial-by-trial basis, Scharlau, Ansorge, and Horstmann (2006) showed that the attentional set adopted to deal with the task at hand affects contingent capture across trials. Taken together, these findings nicely demonstrate that attentional orienting is sensitive to more factors than those manipulated in each trial, and that the processes intervening between cue onset and target response might not be ordered in a strictly serial way. The efficiency with which attention is oriented by directional cues has been shown to be altered also by nonspatial factors, such as duration (Davis & Gibson, 2012), or probability (Girardi et al., 2013) of the SOA. These studies suggest that non-spatial information can influence the voluntary control of spatial attention. Such piece of evidence is important because both current (e.g., Corbetta et al., 2008) and past (e.g., Posner, 1980) theories of attentional control have not considered this possibility. Indeed, research on spatial attention typically assumes that the spatial probability conveyed by a cue elicits voluntary attention only on a trial-by-trial basis. In other words, voluntary control triggered by the cue’s presentation is assumed to be sustained throughout the duration of the SOA, or until the target is delivered. Within this framework, a relevant issue is how spatial cues and non-directional information interact over extended time lags, that is across rather than within trials. In a previous study, we showed that voluntary orienting of attention could be regulated (i.e., switched on/off) within a task according to the relative frequencies of the SOA

Psychological Research

occurrence (Girardi et al., 2013). Time is a crucial factor for the organization of attentional mechanisms. Accordingly, it could be expected to modulate responses also depending on the duration of the within-trial events and on the range of these durations. Here, we directly explored this hypothesis. Since time required to disengage attention and fully re-engage it at a new location has been estimated in 300–650 ms (Cheal, Lyon, & Gottlob, 1994; Duncan, Ward, & Shapiro, 1994; Egeth & Yantis, 1997; Mu¨ller & Rabbitt, 1989; Sperling & Reeves, 1980; Ward, Duncan, & Shapiro, 1996; Yantis, 1998), we used SOA durations within this range. Three equally probable SOAs were arranged within trial blocks in a two-choice RT task. We examined three SOA ranges (250, 350, and 450 ms) centered around two average SOA values (425 and 725 ms) in a factorial design, and assessed the possible effects played on neutral trials (i.e., the FP effect), and on valid vs. invalid trials (i.e., the spatial cueing effect). In other words, by manipulating preparatory demands (average SOA duration), we analyzed FP and spatial cueing effects across several SOA ranges. Novelty of the present study resides in the concurrent manipulation of these two factors. According to literature, the RT-SOA function is expected to become steeper at increasing SOA range (i.e., the larger the spacing of intervals the greater the FP effect) (Elliott, 1973; Gibbon, Church, & Meck, 1984; Karlin, 1959; Niemi & Na¨a¨ta¨nen, 1981). Moreover, these effects should be modulated by the central value of the SOA distribution, being less pronounced when ranges are organized around a higher value (Alegria, 1974; Gallistel & Gibbon, 2000; Grondin & Rammsayer, 2003; Grondin, 2010a; Karlin, 1959; Wearden & Ferrara, 1996). With this in mind, three hypotheses can be put forward as for the effect of temporal context on spatial cueing. 1.

Spatial orienting of attention follows the mandatory routine of disengage/shift/engage operations, which is triggered by cue onset and regulated by the information included in the single trial. If this is the case, responses to validly and invalidly cued targets should align roughly in parallel on the axis of time. Besides, the time course of cueing effects should reveal the orienting profile typical of each SOA (e.g., Mu¨ller & Findlay, 1988; Mu¨ller & Rabbitt, 1989; Nakayama & Mackeben, 1989; Posner & Cohen, 1984). This should hold independently from the other parameters adopted in the study. In particular, since the target appears in one of only two fixed locations, responses to validly and invalidly cued targets are expected to follow a similar function of the SOA changes because the cost of reorienting would reflect a shift of attention covering always the same distance.

2.

Spatial orienting of attention additionally relies on the context of SOAs used in other trials. In this case, the different SOA durations in the whole sequence could produce precise temporal expectancies, with the cue working as a temporal hint in interaction with the spatial information it carries. Accordingly, we should detect some time-based strategy modulating the spatial cueing effect (i.e., the difference between invalid and valid trials) in-line with the experimental manipulation participants were exposed to. In favor of this second hypothesis, robust evidence of a time-based strategy has been already demonstrated with invalid cueing, a condition in which the cognitive effort of investing attentional resources can magnify uncertainty when associated to SOA manipulation (Girardi et al., 2013). In that study, we argued that the mechanism regulating deployment of spatial attention relies on the temporal expectancies implicitly conveyed by the experimental setup. During the experimental session, subjects implicitly learned that a given SOA had more probability to occur. Apparently, they also learned to judge whether enough time had elapsed to justify the possibility that they were facing an invalid cue. Actually, confidence in the cue changes depending on how long the target is delayed: the more the waiting, the more the cue is perceived as invalid. Our results showed that RTs in invalid trials did not include the cost of reorienting that would be expected to add to the duration of the SOA if attention is disengaged after target onset. This suggests that attention was disengaged before target appearance. In other words, a strategy clearly emerged aimed at balancing cognitive resources, i.e., taking as much benefits as possible while trying to minimize the cost of maintaining the focus of attention waiting for a target that may appear elsewhere. Following the same line of thought, in the present experiment we anticipate that responses to invalidly cued targets should not vary according to SOA manipulation. In contrast, responses to validly cued targets should be affected by the increasing SOA durations when the difference between longer and shorter SOA is large. This effect would reflect the alerting components activated by the cue (i.e., the FP effect). We also expect responses to valid trials to be always faster than those to invalidly cued trials since the automatic capture of attention by a target appearing in the cued location should be enough to provide advantages even if attention has been disengaged. This should at least be the case in conditions as used in the present study, with central arrow cues that are predictive of the most likely target location. A similar interplay between top-down and bottom-up mechanisms in a single task would be also in-line with several recent accounts (Petersen & Posner, 2012; Theeuwes, 2010).

123

Psychological Research

3.

The subjective temporal expectation created by the different SOA conditions regulates the engaging rather than the disengaging process. If this was true, participants should engage attention not at the onset but after the offset of the cue, namely while waiting for the target during the SOA. In this case, both responses to invalidly and validly cued targets should be relatively unaffected when SOA duration and range increase.

Methods Participants Sixty students (age 20.98 ± 2.55, 19 males) volunteered for the experiment that was carried out according to the declaration of Helsinki. They were randomly assigned to one of six experimental conditions that differed according to the combination of two between-subjects variable: average stimulus onset asynchrony (SOA) value (2 levels: 425 and 725 ms; low and high, respectively) and SOA range (3 levels: 250, 350, and 450 ms; narrow, middle, and large range, respectively). A schematic layout of the experimental design is depicted in Table 1. All participants were naive as to the purpose of the experiment and had normal or corrected-to-normal vision. Stimuli and procedure Participants sat with their head on a chin rest in front of a computer screen located at a 50-cm viewing distance. A JVC-TK240 camera mounted on the screen was used to monitor fixation. Two boxes (3 wide, 9.6 eccentricity) were located to the right/left of a central fixation point. A central arrow’s head (1.3 eccentricity) acted as a cue as to where a dot or a diamond (1.3 wide) would appear after a variable interval (SOA). Participants had to press one of two vertically aligned keys depending on the target/key association specified for the condition (index finger for diamond and middle finger for dot for one half of the participants, the opposite combination for the other half). All trials had the same structure, and started with the appearance of the cue stimulus for 50 ms followed, after a variable SOA, by the target stimulus that was presented in one of the two boxes. The target remained on the screen until a response was given or 1,500 ms had elapsed. After an interstimulus interval of 500 ms, the next trial began. In the whole sequence, 56 % of trials contained a spatially informative cue, and the target actually appeared on the cued side in 80 % of cases. 34 % of trials were spatially uninformative, being either neutrally cued (bidirectional arrow) or no cued. The

123

10 % of the remaining trials were catch trials including either a single (38 % pointing left, 38 % pointing right) or a bidirectional arrow (24 %) and in which no target appeared. Occurrence of trial types, cued side, and target (diamond/ dot) was randomized within each block. Three SOA ranges (narrow, middle, large) centered around the two average SOA values of 425 (low) or 725 ms (high) were explored. The difference between the longer and the shorter SOA was 250, 350, and 450 ms for the narrow, middle and large SOA range, respectively. Each range included three SOA intervals (short, central, and long). A summary of ranges and duration of the SOAs in each of the six conditions is reported in Table 1. Each group of participants performed 1,278 trials in one experimental session interspersed with 5 pauses. Groups were identified by the ordinal position of the range (narrow: 1, middle: 2, large: 3) followed by a letter that describes the average SOA value (low: L, high: H); the experimental groups were then labeled as 1L, 2L, 3L, 1H, 2H, and 3H.

Results Mean RTs for correct responses and percentages of errors for each group and condition are summarized in Table 2. Less than 2 % of trials were removed from analyses due to anticipation (key press during SOA or RT shorter than 100 ms), miss (no key press or RT longer than 1500 ms) or eye movements. Mean RTs were submitted to a 4 9 3 9 3 9 2 mixedANOVA design with cueing (valid, invalid, neutral, and no cue trials) and SOA (short, central, and long) as repeated measures, and range (narrow, middle, and large) and average SOA value (low, high) as between-subjects factors. Results showed main effects of both cueing, F(3,162) = 276, p \ 0.05, g2p = 0.83, and SOA, F(2,108) = 41.99, p \ 0.05, g2p = 0.44, and no effect of range, F(2,54) = 1.53, p [ 0.1, and average SOA, F \ 1. As for cueing, mean RTs to valid (590 ± 5.11 ms) and invalid (634 ± 5.63 ms) trials were shorter and longer than neutral trials (611 ± 5.35 ms), respectively. Responses to no cued trials (658 ± 5.89 ms) were slower than all other conditions. Considering the SOA effect, responses to trials including the short SOA (631 ± 4.97 ms) were slower compared to trials including the central (620 ± 5.06 ms) and the long SOA (618 ± 5.06 ms) that did not differ from each other. All interactions reached a significance level of p \ 0.05, except range 9 average SOA, F \ 1, cueing 9 SOA range 9 average SOA, p [ 0.2, and cueing 9 SOA 9 average SOA, p [ 0.9. The highest-order interaction of cueing 9 SOA 9 range 9 average SOA (F(12,324) = 3.58, p \ 0.05, g2p = 0.12)

Psychological Research Table 1 Diagram of stimulus onset asynchronies (SOAs) temporal parameters (in ms) according to the experimental groups Range

SOA

Low average SOA (L)

Range

Narrow (250)

Middle (350)

Large (450)

Short

300

250

200

Central

425

425

425

Long

550

600

650

1L

2L

3L

Group

SOA

High average SOA (H) Narrow (250)

Middle (350)

Large (450)

Short

600

550

500

Central

725

725

725

Long

850

900

950

1H

2H

3H

Group

Average SOA refers to the central value of the SOA distribution (low and high, 425 and 725 ms, respectively). Range refers to the amplitude of the difference between the long and the short SOA of each condition

Table 2 Mean reaction times (ms) and error rates (%) (with SE in italics) for valid, neutral, invalid and no cued trials as a function of stimulus onset asynchrony (SOA), SOA range, and average SOA in the experiment Low average SOA SOA (ms)

High average SOA

Valid RT

Err

Neutral

Invalid

RT

RT

Err

Nocue Err

RT

SOA (ms) Err

Valid

Neutral

Invalid

Nocue

RT

Err

RT

Err

RT

Err

RT

Err

577

1.4

587

1.3

608

1.7

651

1.1

16.7

0.5

18.5

0.4

23.2

0.6

24.3

0.4

Central (725)

567

1.4

572

0.4

607

2.5

648

2.5

17.5

0.4

16.2

0.2

17

0.8

22.1

0.8

Long (850)

574

1.2

586

0.8

605

1.4

642

1.9

17

0.3

17.4

0.6

22.4

0.5

23.2

0.6

Narrow SOA range Short (300)

568

1.6

588

2.2

610

2.9

651

1.8

21.1

0.3

23.3

0.91

24.9

0.7

29.1

0.4

Central (425)

575

1.4

592

1.9

604

1.5

636

2.8

24.4

0.3

26.2

0.6

27.9

0.4

28.1

0.8

Long (550)

583

1.2

600

1.7

608

2.5

633

2.1

26.9

0.3

29.5

0.4

29.8

0.9

27.1

0.6

Short (600)

Middle SOA range Short (250) Central (425) Long (600)

608

2.5

622

2.6

624

1.88

634

2.4

26.1

0.5

26.9

0.6

23.9

0.79

26.9

0.5

575

2.4

600

1.4

638

3.13

643

3.2

24.1

0.8

26.4

0.7

30.5

1.29

28.3

0.8

578

2

597

2.4

628

2.92

637

2.1

24.7

0.3

25.9

0.6

25.6

1.21

26.8

0.6

Short (550) Central (725) Long (900)

618

1.8

631

2.8

633

2.5

663

2.6

19.5

0.5

17

0.9

18.7

0.7

21

0.8

593

1.4

616

1.9

640

2.08

662

2.9

18.2

0.3

17.9

0.5

17.9

0.7

22.9

0.6

593

1.4

614

1.8

651

1.9

670

2.8

16.7

0.5

16

0.5

18.2

0.7

21.6

0.6

628

2.2

650

2.6

657

1.9

683

1.8

30.3

0.7

30

1

27.9

1

33.3

1.2

Large SOA range Short (200) Central (425) Long (650)

637

2.2

666

2.4

681

4.4

678

2.8

14.8

0.7

14.9

1.1

16.8

1.2

16.1

0.5

590

1.8

627

1.9

660

3.5

668

1.7

13.1

0.6

13.9

0.5

16.2

1.3

13.6

0.4

559

1.6

595

2.8

653

2.9

671

1.9

12.6

0.3

13.6

0.7

19.9

0.8

16.6

0.7

was specifically evaluated in three ways (see Fig. 1). First, a series of pairwise comparisons was run to assess how the pattern of RTs varied within each group. Pairwise comparisons were applied using Bonferroni’s correction. A type I error rate of a = 0.05 was used in all statistical tests. Given a = 0.05 and n = 10 participants an effect of size dz = 0.9 (cf. Cohen, 1977) could be detected with a probability of 1 - b = 0.84.1 Second, ANOVA trend tests were carried out 1

All statistical power analyses reported here were performed using the GPOWER program (Faul, Erdfelder, Lang, & Buchner, 2007).

Short (500) Central (725) Long (950)

604

1.5

628

1.8

649

1.9

687

2.5

29

0.4

30.4

0.7

31.7

0.7

33.8

0.4

597

1.7

622

1.7

655

1.5

693

3.3

25.7

0.6

27.3

0.6

29.4

0.9

33.1

1.3

on neutral trials to examine the FP effect, namely how RTs changed across SOAs in the different combinations of SOA range and average SOA. The hypothesis of both linear and quadratic trends of RTs across SOAs was analyzed within each group. Third, to specifically address the effect of spatial cueing across experimental conditions we considered the spatial cueing effect (i.e., the mean difference between RTs to invalid and valid trials computed for each participant) as dependent measure and ran a three-way mixed-model ANOVA with SOA range and average SOA as betweensubjects factors and SOA as repeated factor.

123

Psychological Research

valid

neutral

nocue

Middle Range

Large Range

(Group 1H)

(Group 2H)

(Group 3H)

(Group 1L)

(Group 2L)

(Group 3L)

Narrow Range 700

invalid

680 660 640

Reacon Times (msec)

620 600 580 560

High Average SOA 700 680 660 640 620 600 580 560 short

central

Low Average SOA

long

short

central

long

short

central

long

Smulus Onset Asynchrony (SOA)

Fig. 1 Mean reaction times (ms) for valid, neutral, invalid, and no cued targets as a function of the short, central and long SOA in each group. The six experimental groups are depicted according to the average SOA value (high and low: upper and lower panels,

respectively) and SOA range (narrow, middle and large: left, center and right panels, respectively). Note the peculiar flattening of RTs to invalid and no cued targets in all conditions (see Text for details)

Results from pairwise comparisons, ANOVA trend tests and spatial cueing effects are separately detailed.

Within group 1H (1 = narrow range, H = high average SOA) the effect of alerting (64, 75, and 56 ms for the short, central, and long SOA, respectively) was significant with all SOAs. No significant benefit was detected albeit a slight difference emerged between RTs to validly and neutrally cued targets (10, 5, and 12 ms). Significant costs for reorienting after invalidly cued targets (21, 35, and 20 ms) were found with all SOAs. Pairwise comparisons on individual SOA levels showed a lack of differences between RTs to invalidly and no cued target, while RTs to neutrally and validly cued targets were significantly faster with the central compared to the shorter SOA (see Fig. 1). In group 2L and 2H (2 = middle range, L, H = low and high average SOA), the effects of alerting, benefits, and costs were significant in trials with the central (group 2L: 43, 25, 38 ms, and group 2H: 46, 23, 23 ms, respectively) and long SOA (group 2L: 39, 19, 31 ms, and group 2H: 55, 21, 37 ms). In the condition with the short SOA, an effect of alerting was found significant only on RTs of group 2H (32 ms; group 2L: 12 ms). Cueing produced benefits in both groups (14 and 13 ms, respectively). Costs were not significant (2 ms for both 2L and 2H groups) (see Table 3). Further comparisons between SOA levels for each cueing

RTs within groups comparisons Pairwise comparisons (see Table 3 for t values) showed that within group 1L (1 = narrow range, L = low average SOA) RTs in trials including the short SOA were significantly slower for no cued targets compared to neutrally cued targets, revealing a significant alerting effect (63 ms). RTs to validly cued targets were significantly faster than RTs to neutrally cued targets, indicating a significant benefit effect (20 ms). RTs to invalidly cued targets were significantly slower than RTs to neutrally cued targets revealing a cost associated with reorienting attention (22 ms). Conversely, in trials with central and long SOAs only a significant alerting effect (43 and 33 ms, respectively) and a benefit from valid cueing (17 and 16 ms, respectively) were found. The costs in invalid compared to neutral conditions failed to reach significance (12 and 8 ms with central and long SOA, respectively). No difference emerged between the single SOAs in any cueing condition (see Fig. 1).

123

Psychological Research Table 3 Pairwise comparisons between responses to no cued vs. neutral trials (alerting), neutral vs. valid trials (benefits) and neutral vs. invalid trials (costs) according to group (1 = narrow, 2 = middle, and 3 = large SOA range; L = low, H = high average SOA), and SOA duration (short, central and long)

SOA

Group 1L

Group 1H

Group 2L

Group 2H

Group 3L

Each cell reports the magnitude of cueing effect (in ms), and the corresponding t and dz values (9 df) All values significant (p \ 0.05) unless otherwise indicated (x)

Group 3H

Short Central Long Short Central Long Short Central Long Short Central Long Short Central Long Short Central Long

Alerting

Benefits

Costs

ms

t

dz

ms

t

dz

ms

t

dz

63 43 33 64 75 56 12 43 39 32 46 55 12 41 77 34 59 71

7.34 9.38 3.82 6.29 10.42 6.89 2.14x 8.75 4.78 5.96 5.02 5.44 4.3 9.8 12.58 5.35 8.32 10.03

2.32 2.97 1.21 2.0 3.26 2.15 \0.6 2.87 1.5 1.88 1.64 1.72 1.36 3.11 3.98 1.7 2.57 3.23

20 17 16 10 5 12 14 25 19 13 23 21 28 37 36 21 24 24

5.45 4.55 3.15 2.29x \1x 2.7x 3.71 8.5 4.28 2.68 7.52 9.31 8.7 9.77 13.51 8.56 8.59 8.92

1.7 1.43 0.99 0.7 0.22 0.8 1.17 2.69 1.35 0.87 2.35 3.0 2.75 3.09 4.27 2.62 2.67 2.82

22 12 8 21 35 20 2 38 31 2 23 37 15 33 59 7 22 33

4.72 3.98 1.5x 3.02 10.95 2.05x \1x 5.7 3.1 \1x 3.16 5.11 6.14 8.65 7.19 1.29x 10.69 8.35

1.49 1.3 \0.06 0.96 3.5 \0.6 \0.05 1.8 0.98 \0.08 1.0 1.61 1.94 2.73 2.27 0.38 3.67 2.54

condition revealed that RTs to validly and neutrally cued targets were slower at the short SOA compared to the other two levels. No difference was detected between invalidly or no cued targets across all SOAs (see Fig. 1). As for group 3L (3 = large range, L = low average SOA), RTs to trials with both the central and long SOA produced significant effects of alerting, benefits, and costs (41, 37, 33 ms, and 77, 36, 59 ms, respectively). With the short SOA, only benefits (28 ms) were significant (alerting: 12 ms, and costs: 15 ms). In this group, RTs to both neutral and valid cueing were significantly slower across the three SOA levels (short [ central [ long SOA) while responses to invalidly cued targets were slower only at the short SOA compared to the other two levels. RTs to no cued targets did not differ across all SOAs. Finally, within group 3H (3 = large range, H = high average SOA), RTs to trials with both the central and long SOA showed significant effects of alerting, benefits, and costs (59, 24, 22 ms, and 71, 24, 33 ms, respectively). With the short SOA only the effects of alerting (34 ms) and benefits (21 ms) were significant (costs: 7 ms). Again, RTs to validly and neutrally cued targets were slower at the short SOA compared to the two other SOAs (which not differed from each other); no difference emerged according to SOA levels regarding invalidly or no cued targets (see Fig. 1). Errors A 4 9 3 9 3 9 2 mixed-ANOVA design with cueing (valid, invalid, neutral, and no cue trials) and SOA (short, central, and long) as repeated measures, and range (narrow,

middle, and large) and average SOA value (low, high) as between-subjects factors was run on the arcsine square root transforms of percentages of incorrect responses. No main effect of cueing, F(3,162) = 1.25, p [ 0.2, SOA, F \ 1, range, F \ 1, and average SOA, F(1,54) = 1.54, p [ 0.2, emerged (see Table 2 for error values). Considering interactions, cueing 9 SOA (F(6,324) = 2.85, p \ 0.05) was significant due to a higher error rate at the central SOA in no cue trials as compared to neutral trials. Error rates were not different in cueing 9 range, F(6,162) = 1.29, p [ 0.2, cueing 9 average SOA, F(3,162) = 1.88, p [ 0.1, cueing 9 range 9 average SOA, F(6,162) = 1.86, p [ 0.09, SOA 9 average SOA, F(2,108) = 1.17, p [ 0.3. SOA 9 range, average SOA 9 range, SOA 9 range 9 average SOA, cueing 9 SOA 9 range, cueing 9 SOA 9 average SOA (all Fs \ 1). FP effect As can be seen in Fig. 2, a linear trend linking the decrease of RT to the increase of SOA durations for the different SOA ranges was found significant only in group 2L and 3L (F(1,9) = 16.28, p \ 0.05, g2p = 0.64 and F(1,9) = 270.28, p \ 0.05, g2p = 0.97, respectively; group 1L: F(1,9) = 1.84, p [ 0.1). The test for the quadratic trend was not significant (F(1,9) = 3.96, p [ 0.08; F(1,9) = 3.73, p [ 0.09; and F \ 1, respectively). Likewise, in the high average SOA value condition, the same linear trend was found significant in group 2H and 3H (F(1,9) = 45.89, p \ 0.05, g2p = 0.84; and F(1,9) = 49.62, p \ 0.05, g2p = 0.85 respectively; group 1H F \ 1). The

123

Psychological Research

Reaction Times (msec)

680

Low

Spatial cueing effect

High

660 640 620 600 580 560

0

200

400

600

800

1000

SOAs Group 1L Group 2L Group 3L

Group 1H Group 2H Group 3H

Fig. 2 Mean reaction times (ms) in neutral trials as a function of stimulus onset asynchrony (SOA). Low and high refer to the average SOA value. Groups are depicted according to the SOA ranges (1 = narrow, 2 = middle, and 3 = large). The foreperiod (FP) effect changes according to SOA parameters. The FP effect increases with greater SOA ranges (with constant average SOA) and decreases with greater average SOA (with constant SOA range)

quadratic trend was also significant (F(1,9) = 5.67, p \ 0.05; F(1,9) = 5.16, p \ 0.05, g2p = 0.36; F(1,9) = 10.62, p \ 0.05, g2p = 0.54 group 1H, 2H, and 3H, respectively). Additional comparisons showed that the linear trend was larger both in group 3L (as compared to group 2L, t(18) = 6.11, p \ 0.05, d = 2.88) and 3H (as compared to group 2H, t(18) = 2.4, p \ 0.05, d = 1.13); no difference emerged for the quadratic trends, t \ 1. In other words, the FP effect increased with the SOA range when the average SOA value was held constant. Furthermore, the linear trend was larger in group 3L than in group 3H (t(18) = 7.37, p \ 0.05, d = 3.47), while no difference was found between groups 2L and 2H (t(18) = 1.23, p [ 0.2). In other words, in the groups exposed to the larger SOA range, the FP effect decreased in participants presented with the high average SOA.2 2

In order to estimate the contribution of automatic sequential effects to the FP effect (see Los & Van den Heuvel, 2001), we made an analysis of the experimental findings that included the current SOA (short, central, long), the SOA of the previous trial, i.e., SOAn - 1 (short, central, long, catch), and cueing (valid, invalid, neutral, no cue) as within-subjects factors, and average SOA and SOA range as between-subjects variables. The ANOVA conducted on mean RTs revealed no contribution of SOAn - 1. SOAn - 1 had neither a main effect, F(3,108) = 1.69, nor did it interact with SOA in any case: SOAn – 1 9 SOA, F(6,216) = 1; SOAn – 1 9 SOA 9 range 9 average SOA, F \ 1. Therefore, in this task, asymmetric sequential effects (see Los & Van den Heuvel, 2001), did not contribute to the foreperiod effect.

123

The three-way mixed-model ANOVA on RTs with SOA range (narrow, middle, large) and average SOA (low, high) as between-subjects factors, and SOA (short, central, long) as repeated factor, showed significant main effects of SOA range, F(1,54) = 5.61, p \ 0.05, g2p = 0.17, and SOA, F(2,108) = 28.92, p \ 0.05, g2p = 0.35 (average SOA, F(1,54) = 2.23, p [ 0.1). As for SOA range, the cueing effect was greater in the large range (57 ± 5.03 ms) compared to both the narrow (33 ± 5.17 ms) and the middle range (41 ± 5.51 ms) that did not differ from each other. Concerning SOA, the cueing effect was smaller with the short (29 ± 3.72 ms) than the central (49 ± 3.39 ms) and the long SOA (53 ± 4.73 ms) that did not differ from each other. The SOA range 9 SOA interaction was significant, F(4,108) = 13.95, p \ 0.05, g2p = 0.34. In the large range, the cueing effect differed significantly among all SOAs (short SOA = 36 ± 4.5 ms, central SOA = 58 ± 4.97 ms, and long SOA = 76 ± 7.4 ms). In the middle range condition, the cueing effect was smaller with the short (15 ± 6.98 ms) as compared to both the central (55 ± 6.33 ms) and the long SOA (54 ± 7.43 ms). In the narrow range, the magnitude of the cueing effect did not differ among SOAs (37 ± 6.66, 34 ± 5.05 and 28 ± 6.07 ms for the short, the central and the long SOA, respectively) (all comparisons, p \ 0.05, with Bonferroni’s correction). The three-way interaction between SOA, SOA range and average SOA, was also significant, F(4,108) = 3.08, p \ 0.05, g2p = 0.1. In the low average SOA condition, the following emerged. With the short SOA, the cueing effect did not differ among SOA ranges (group 1L = 42 ± 7.81 ms, group 2L = 16 ± 10.79 ms, and group 3L = 44 ± 4.63 ms). With the central SOA, the cueing effect was smaller in group 1L (29 ± 5.27 ms) as compared to both groups 2L (63 ± 9.08 ms) and 3L (70 ± 7.31 ms), which did not differ from each other. Finally, with the long SOA the cueing effect was greater in group 3L (95 ± 10.73 ms) as compared to both group 1L (24 ± 6.6 ms) and 2L (50 ± 12.55 ms), that did not differ from each other. In the high average SOA, the cueing effect did not differ among SOAs and SOA ranges (31 ± 10.94, 40 ± 8.55, and 32 ± 10.45 ms in group 1H, 14 ± 9.45, 46 ± 8.49, and 58 ± 8.5 ms, in group 2H, and 29 ± 7.18, 46 ± 4.32, and 58 ± 6.35 ms, in group 3H for the short, the central and the long SOA, respectively) (all comparisons p \ 0.05, with Bonferroni’s correction). SOA range 9 average SOA, and SOA 9 average SOA were not significant (F(2,54) = 2.08, p [ 0.1, and F \ 1, respectively). To sum up, results showed that the magnitude of spatial cueing varied with average SOA and SOA range.

Psychological Research

Specifically, the cueing effect was greater if average SOA was low and SOA range was large (group 3L) (compared to middle or narrow, i.e., groups 1L and 2L), particularly for the long SOA. In the case of the central SOA, the cueing effect was smaller if the SOA range was narrow (group 1L) (compared to middle and large, groups 2L and 3L). No such differences emerged for high average SOA.

Discussion In a spatial cueing task, by manipulating the range and the average value of SOA distribution, we observed a peculiar modulation of responses according to both the alerting and spatial information provided by the cue. This modulation was particularly evident when the average SOA was low (425 ms). Namely, in conditions applying middle and large SOA ranges (350 and 450 ms of difference between the longest and the shortest SOA, respectively), no RT difference emerged when neutrally cued targets were compared to both no cued and invalidly cued targets (i.e., there were no alerting effects and no spatial costs). This was true for the shortest SOA durations tested. Conversely, in conditions applying the narrow SOA range (range 1 = 250 ms), spatial costs were still seemingly suppressed but this was true only for the longest SOA duration. In addition, the difference between RTs to neutrally and validly cued targets (i.e., spatial benefits) was significant. These observations clearly indicate that responses were systematically influenced by the context provided by the timing of cue-target SOAs. Responses to neutrally cued targets showed that preparedness improved proportionally to the increasing of SOA (i.e., the FP effect). The decrease of RT as a function of the SOA is a well-known preparatory effect and it is attributed to the growth of expectancy as time runs out without target presentation (Elithorn & Lawrence, 1955; Niemi & Na¨a¨ta¨nen, 1981; Sanders, 1966). As depicted in Fig. 2, in the present case RTs in the narrowest SOA range condition (i.e., groups 1L and 1H) did not show this effect of preparation over time. In contrast, the FP effect was clearly evident when SOA differences were spaced across a large range. Indeed, increasing the spacing between SOAs enhances their discriminability, especially if they are distributed across a larger range of durations (Elliott, 1973; Grondin, & Rammsayer, 2003; Karlin, 1959). The present findings further show that participants’ response is modulated by cue-target SOAs both within (i.e., SOA duration) and between (i.e., SOA range) trials. This is consistent with psychophysics (e.g., Grondin, 2001, 2010b; Wearden & Lejeune, 2008) and reaction time reports (e.g., Alegria, 1974; Karlin, 1959) and is thought to reflect the reduced ability of participants to estimate durations when their

scaling shortens (Correa, Lupia´n˜ez, & Tudela, 2005; Gallistel & Gibbon, 2000; Klemmer, 1956; Steinborn, Rolke, Bratzke, & Ulrich, 2008). The most interesting aspect of our results is that effectiveness of the spatial cue differed as a function of temporal context. Specifically, effectiveness of the valid cue increased rapidly as SOA increased from short to long, with a peak that varied for each SOA depending on the range. With shorter SOAs, valid cues produced benefits in all conditions of SOA range (narrow, middle, large). With longer SOAs, they produced benefits only in the large SOA range condition (see Fig. 1). On the contrary, RTs to invalidly cued targets were not similarly modulated by either SOA duration or range (see Fig. 1). This result suggests that at short intervals attention is drawn exogenously to the location indicated by cue (Cheal & Lyon, 1991; Jonides, 1981). At larger SOA ranges, longer SOAs last a sufficient time to allow endogenous attention mechanisms to intervene (Posner, 1980; Cheal et al., 1994; Egeth & Yantis, 1997; Ward et al., 1996; Yantis, 1998). Endogenous spatial cueing has been previously described as reflecting the disengaging/shifting/engaging operations, and taking about 300–650 ms to reach its maximal efficacy (Mu¨ller & Findlay, 1988; Mu¨ller & Rabbitt, 1989; Posner & Cohen, 1984; Posner et al., 1987). Based on these findings, we assume that the observed benefit modulation by SOA duration and range is due to the further recruitment of voluntary orienting to the cued location. According to literature on spatial attention (Mu¨ller & Rabbitt, 1989; Nakayama & Mackeben, 1989; Posner & Cohen, 1984; Posner et al., 1987; Ward et al., 1996), for both valid and invalid trials, response time distributions as a function of SOA are expected to be the same. In addition, their difference (i.e., the cost of reorienting) is expected to remain stable across conditions if shifting of attention occurs between two fixed positions (as required by the present task). As reported in Fig. 1, our data clearly show that this is not the case. Apparently, responses in invalid trials are not beset with the same reorienting costs in all conditions. In fact, the corresponding influence of SOAs on invalid RTs is mitigated in comparison to the influence of SOAs on valid (but also neutral) trials. The possibility exists that the pattern of spatial costs found here may reflect a ceiling effect on RTs due to reduced alertness. However, this account would not explain the greater slowing down observed with no cued compared to invalid trials (see Fig. 1). Moreover, participants tested on the low average SOA value were more affected by SOA range. Namely, the largest range of SOAs intensified the effects of the FP effect, spatial costs and benefits recorded at the longer SOA (Fig. 1, lower right panel). It should be noted that according to the interpretation of cue-triggered mechanisms of disengaging–shifting–engaging operations, the

123

Psychological Research

cost of reorienting is expected to be greater with the narrowest range of SOAs and smaller with the largest range of SOAs (Posner, 1980). This would imply that, on the one hand, participants responding to trials including longest SOA should have smaller costs since they are allowed a greater amount of time to disengage attention following an invalid cue (Mu¨ller & Rabbitt, 1989; Posner et al., 1987). On the other hand, participants with the narrowest range of SOAs should show the largest costs because of the relatively less time to disengage and shift attention before responding (Mu¨ller & Rabbitt, 1989; Posner et al., 1987). Here, we found quite the opposite: the smaller the SOA range, the smaller the costs (see Fig. 1, lower left and right panels). The effect of the temporal context is also evident if we consider the spatial cueing effect (i.e., the invalid/valid RT difference) that is clearly modulated by both SOA duration and range. The central SOA produced a different effect in the narrow as compared to the middle and large SOA ranges. This is a significant demonstration that the directional cueing of attention is context sensitive because the same trial can elicit a different pattern of response depending on the sequence in which it is inserted. Taken together, our results challenge the broadly accepted assumption that, within each trial, selective attending follows the mandatory routine of disengage/shift/ engage (Posner et al., 1987) triggered by the cue and temporally marked by target onset. Such a routine implies that compared to valid trials, responses in invalid trials should maintain the same difference as a function of SOA because of the constant extra amount of time required to perform, in all conditions, the cognitive operations of disengaging, shifting and re-engaging of attention. In other words, data should show a similar time course of the SOA/ RT function across all SOA parameters. The peculiar pattern shown here by costs of invalid cueing rather indicates some change in strategy that occurs within the block of trials and depends on the SOA range presented. A possible explanation for the diverging shapes of RTs in valid and invalid trials could be that subjects attended to the cued location but disengaged attention before target onset. In this case, validly cued responses could still benefit from the automatic orienting of attention elicited by the directional component of the cue (Tipples, 2002), leading to prompt detection of the target on the cued side. An early disengagement can also explain why, in invalid trials, the cost of reorienting to the unattended location is unaffected by SOA duration. Actually, if disengagement and shifting of attention occur only after target appearance, the corresponding RT should necessarily include the time required by these operations plus the entire duration of the SOA. Our results show that RTs in invalid trials did not change as a function of SOA, especially with the larger ranges (see Fig. 1).

123

Since the cue also provides a crucial warning/alerting information (Posner & Boies, 1971), we maintain that time (namely, the interval elapsed since cue onset) is the pivot key allowing the brain to reach a status that we could define as of ‘uncertainty about the cue’. This point of uncertainty should represent the moment in which subjective expectancy about where the target will effectively appear is equivalent for both the cued and the uncued location. In other words, since subjects have been cued towards one side and time runs but nothing occurs, the possibility that the unattended location will host the target strengthens while the perceived probability of having received a valid cue weakens. In-line with previous results (Girardi et al., 2013), the inter-trial temporal context (i.e., the different SOA durations in the whole sequence of trials) provides useful information and allows the creation of a precise temporal expectancy. Namely, it identifies a reference interval that marks the point at which in most trials something has to happen. We consider this point (in time) as the cognitive marker for the subjective cue uncertainty. At that moment, disengaging attention from the cued side would be the best way to adapt to the new subjective probabilities according to which the two locations are now perceived as potentially equivalent. It should be noted that, since we included 10 % of catch trials in the task (namely, conditions in which the target never appeared in spite of cueing), the supposed advantage would exist only in disengaging attention from the formerly cued location and not in shifting it to attend to the opposite side. Lacking a further engagement, the cued location should, however, maintain a relative salience and the time required for the eventual reorienting should not vary across conditions. This pattern fits exactly with our results: responses in valid trials are faster and the cost of reorienting in invalid trials does not follow the SOA duration changes, as expected if disengagement of attention were performed before target onset. An alternative account could be that attention is withheld at cue onset and is shifted only at the end of the cue-target interval (see e.g., Yantis, 1988). This possibility would be consistent with the reported pattern of costs in invalid trials but not with the advantages observed in valid trials in relation to SOAs changes. Indeed RTs in valid trials shortened at the lengthening of the SOA, meaning that attention had been engaged to the cued location before target onset. It is interesting to note that in the literature, relevance of the temporal context in organizing cognitive activity has proved already evident at single cell level. Neurons process elapsing of time as a function of the probability of an attended event (Janssen & Shadlen, 2005). These data are a clear indication that temporal information is included in the functional repertory of the nervous system down to the level of the working cell. Indeed, from an ecological perspective the advantage of anticipating the occurrence of

Psychological Research

predictable events is self-evident, even on a scale of seconds or less. More to the point, the anticipation of behaviorally relevant events has been shown to occur following a strategy based on task’s temporal attributes to which the brain is not explicitly cued nor incidentally required to monitor (Ghose & Maunsell, 2002). In-line with these observations, we consider the fact that the average RTs were not altered by SOAs in invalid trials as a clear demonstration of some economical strategy. Spending resources to concentrate attention (endogenously) according to the spatial suggestion would be profitable only within a relatively reduced temporal frame. Otherwise, switching-off selective attention is intended to spare cognitive resources in a context that does not warrant for an easy and secure gain. In other words, SOAs may lengthen but attentional costs are controlled so that waiting for a target never exceeds a certain amount of cognitive effort. The interpretation we propose is based on the possibility that the task involves voluntary control only until subjects are actively focusing on the cued location. When temporal parameters enhance the perceived probability that the target appears outside the attentional window, voluntary control drops. Switching-off of selective attention allows for an involuntary process to take place, and permits target detection through automatic capture (see also Girardi et al., 2013). A similar account, that describes selective attending as resulting from the dynamic interplay of distinct attentive modalities fits with recent literature which considers attention within the broader frame of a complex cognitive network where several systems work in interaction both in orienting and executive control (Petersen & Posner, 2012). In conclusion, our data show that attentional shifting is more context sensitive than expected. Cognitive operations usually ascribed to within-trial processes appear to respond to a broad set of information that go beyond the level of single events and include their relations on a wider scale (Dosenbach, Fair, Cohen, Schlaggar, & Petersen, 2008). In this view, time is a crucial factor. The question remains whether the cognitive system responds to time per se or else evaluates the biological effort produced by certain operations and this, in turn, contributes to temporal sensitivity (Gorea, 2011). Acknowledgments The work was supported by a grant from the Compagnia San Paolo di Torino, Programma Neuroscienze 2008/09 (TECRONE-project). The authors wish to thank two anonymous reviewers for their valuable comments and Dr. Elena Daprati for her suggestions and assistance in the preparation of the manuscript.

References Alegria, J. (1974). The time course of preparation after a first peak: Some constraints of reacting mechanisms. The Quarterly Journal of Experimental Psychology, 26, 622–632.

Alegria, J., & Bertelson, P. (1970). Time uncertainty, number of alternatives and particular signal-response pair as determinants of choice reaction time. Acta Psychologica, 33, 36–44. Cheal, M. L., & Lyon, D. R. (1991). Central and peripheral precuing of forced-choice discrimination. The Quarterly Journal of Experimental Psychology, 43A, 859–880. Cheal, M. L., Lyon, D. R., & Gottlob, L. R. (1994). A framework for understanding the allocation of attention in location-precued discrimination. The Quarterly Journal of Experimental Psychology, 47, 699–739. Cohen, J. (1977). Statistical power analysis for the behavioral sciences. Hillsdale: Lawrence Erlbaum Associates. Corbetta, M., Patel, G., & Shulman, G. L. (2008). The reorienting system of the human brain: from environment to theory of mind. Neuron, 58(3), 306–324. Correa, A., Lupia´n˜ez, J., & Tudela, P. (2005). Attentional preparation based on temporal expectancy modulates processing at the perceptual-level. Psychonomic Bulletin & Review, 12(2), 328–334. Cui, X., Stetson, C., Montague, P. R., & Eagleman, D. M. (2009). Ready…go: Amplitude of the fMRI signal encodes expectation of cue arrival time. PLoS Biology, 7(8), e1000167. Davis, G. J., & Gibson, B. S. (2012). Going rogue in the spatial cuing paradigm: High spatial validity is insufficient to elicit voluntary shifts of attention. Journal of Experimental Psychology: Human Perception and Performance, 38(5), 1192–1201. Dosenbach, N. U. F., Fair, D. A., Cohen, A. L., Schlaggar, B. L., & Petersen, S. E. (2008). A dual-networks architecture of top-down control. Trends in Cognitive Sciences, 12(3), 99–105. Drazin, D. H. (1961). Effects of foreperiod, foreperiod variability, and probability of stimulus occurrence on simple reaction time. Journal of Experimental Psychology, 62, 43–50. Duncan, J., Ward, R., & Shapiro, K. (1994). Direct measurement of attentional dwell time in human vision. Nature, 369, 313–315. Egeth, H. E., & Yantis, S. (1997). Visual attention: control, representation, and time course. Annual Review of Psychology, 48, 269–297. Elithorn, A., & Lawrence, C. (1955). Central inhibition: Some refractory observations. Quarterly Journal of Experimental Psychology, 11, 211–220. Elliott, R. (1973). Some confounding factors in the study of preparatory set in reaction time. Memory and Cognition, 1, 13–18. Eriksen, C. W., & St. James, J. D. (1986). Visual attention within and around the field of focal attention: a zoom lens model. Perception and Psychophysics, 40, 225–240. Fan, J., McCandliss, B. D., Sommer, T., Raz, A., & Posner, M. I. (2002). Testing the efficiency and independence of attentional networks. Journal of Cognitive Neuroscience, 14, 340–347. Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39, 175–191. Gabay, S., & Henik, A. (2010). Temporal expectancy modulates inhibition of return in a discrimination task. Psychonomic Bulletin and Review, 17(1), 47–51. Gallistel, C. R., & Gibbon, J. (2000). Time, rate, and conditioning. Psychological Review, 107(2), 289–344. Ghose, G. M., & Maunsell, J. H. R. (2002). Attentional modulation in visual cortex depends on task timing. Nature, 419, 616–620. Gibbon, J., Church, R. M., & Meck, W. H. (1984). Scalar timing in memory. In J. Gibbon & L. G. Allan (Eds.), Annals of the New York Academy of Sciences, 423, Timing and time perception (pp. 52–77). New York: New York Academy of Sciences. Girardi, G., Antonucci, G., & Nico, D. (2013). Cueing spatial attention through timing and probability. Cortex, 49(1), 211–221.

123

Psychological Research Gorea, A. (2011). Ticks per thought or thoughts per tick? A selective review of time perception with hints on future research. Journal of Physiology-Paris, 105, 153–163. Grondin, S. (2001). From physical time to the first and second moments of psychological time. Psychological Bulletin, 127(1), 22–44. Grondin, S. (2010a). Timing and time perception: A review of recent behavioral and neuroscience findings and theoretical directions. Attention, Perception, & Psychophysics, 72(3), 561–582. Grondin, S. (2010b). Unequal Weber fractions for the categorization of brief temporal intervals. Attention, Perception, & Psychophysics, 72(5), 1422–1430. Grondin, S., & Rammsayer, T. (2003). Variable foreperiods and temporal discrimination. The Quarterly Journal of Experimental Psychology, 56A, 731–765. Hommel, B., Pratt, J., Colzato, L., & Godijn, R. (2001). Symbolic control of visual attention. Psychological Science, 12(5), 360–365. Janssen, P., & Shadlen, M. N. (2005). A representation of the hazard rate of elapsed time in macaque area LIP. Nature Neuroscience, 8(2), 234–241. Jongen, E. M. M., & Smulders, F. T. Y. (2007). Sequence effects in a spatial cueing task: endogenous orienting is sensitive to orienting in the preceding trial. Psychological Research, 71(5), 516–523. Jonides, J. (1981). Voluntary versus automatic control over the mind’s eyes. In J. Long & A. Baddeley (Eds.), Attention and performance IX (pp. 187–203). Hillsdale: Erlbaum. Karlin, L. (1959). Reaction time as a function of foreperiod duration and variability. Journal of Experimental Psychology, 58, 185–191. Klein, R., & Hansen, E. (1990). Chronometric analysis of apparent spotlight failure in endogenous visual orienting. Journal of Experimental Psychology: Human Perception and Performance, 16(4), 790–801. Klemmer, E. T. (1956). Time uncertainty in simple reaction time. Journal of Experimental Psychology, 51, 179–184. Lambert, A., Naikar, N., McLachlan, K., & Aitken, V. (1999). A new component of visual orienting: Implicit effects of peripheral information and subthreshold cues on covert attention. Journal of Experimental Psychology: Human Perception and Performance, 25, 321–340. Los, S. A., & Schut, M. L. J. (2008). The effective time course of preparation. Cognitive Psychology, 57, 20–55. Los, S. A., & Van Den Heuvel, C. E. (2001). Intentional and unintentional contributions of nonspecific preparation during reaction time foreperiods. Journal of Experimental Psychology: Human Perception and Performance, 27(2), 370–386. Mu¨ller, H. J., & Findlay, J. M. (1988). The effect of visual attention on peripheral discrimination thresholds in single and multiple element displays. Acta Psychologica, 69, 129–155. Mu¨ller, H. J., & Rabbitt, P. M. A. (1989). Reflexive and voluntary orienting of visual attention: Time course of activation and resistance to interruption. Journal of Experimental Psychology: Human Perception and Performance, 15, 315–330. Mu¨ller-Gethmann, H., Ulrich, R., & Rinkenauer, G. (2003). Locus of the effect of temporal preparation: evidence from the lateralized readiness potential. Psychophysiology, 40, 597–611. Na¨a¨ta¨nen, R. (1970). The diminishing time-uncertainty with the lapse of time after the warning-signal in reaction time experiments with varying fore-periods. Acta Psychologica, 34, 399–419. Na¨a¨ta¨nen, R. (1972). Time uncertainty and occurrence uncertainty of the stimulus in a simple reaction time task. Acta Psychologica, 36, 492–503.

123

Nakayama, K. (1990). The iconic bottleneck and the tenuous link between early visual processing and perception. In C. Blakemore (Ed.), Vision: Coding and efficiency (pp. 411–422). Cambridge: Cambridge University Press. Nakayama, K., & Mackeben, M. (1989). Sustained and transient components of focal visual attention. Vision Research, 29, 1631–1647. Niemi, P., & Na¨a¨ta¨nen, R. (1981). Foreperiod and simple reaction time. Psychological Bulletin, 89, 133–162. Nobre, A. C., Correa, A., & Coull, J. T. (2007). The hazards of time. Current Opinion in Neurobiology, 17, 1–6. Petersen, S. E., & Posner, M. I. (2012). The attention system of the human brain: 20 years after. Annual Review of Neuroscience, 35, 73–89. Posner, M. I. (1980). Orienting of attention. The Quarterly Journal of Experimental Psychology, 32, 3–25. Posner, M. I., & Boies, S. J. (1971). Components of attention. Psychological Review, 78, 391–408. Posner, M. I., & Cohen, Y. (1984). Components of visual orienting. In H. Bouma & D. G. Bouwhuis (Eds.), Attention and performance X (pp. 531–556). Hillsdale: Erlbaum. Posner, M. I., Cohen, A., & Rafal, R. D. (1982). Neural systems control of spatial orienting. Proceedings of the Royal Society of London, B, 298, 187–198. Posner, M. I., Inhoff, A. W., Friedrich, F. J., & Cohen, A. (1987). Isolating attentional systems: a cognitive-anatomical analysis. Psychobiology, 15, 107–121. Posner, M. I., & Petersen, S. E. (1990). The attention system of the human brain. Annual Review of Neuroscience, 13, 25–42. Posner, M. I., Petersen, S. E., Fox, P., & Raichle, M. E. (1988). Localization of cognitive operations in the human brain. Science, 240, 1627–1631. Posner, M. I., Snyder, C. R. R., & Davidson, B. J. (1980). Attention and detection of signals. Journal of Experimental Psychology: General, 109(2), 160–174. Ranzini, M., Dehaene, S., Piazza, M., & Hubbard, E. M. (2009). Neural mechanisms of attentional shifts due to irrelevant spatial and numerical cues. Neuropsychologia, 47(12), 2615–2624. Raz, A., & Buhle, J. (2006). Typologies of attentional networks. Nature Reviews Neuroscience, 7, 367–379. Sanders, A. F. (1966). Expectancy: application and measurement. Acta Psychologica, 25, 293–313. Scharlau, I., Ansorge, U., & Horstmann, G. (2006). Latency facilitation in temporal-order judgments: time course of facilitation as a function of judgment type. Acta Psychologica, 122, 129–159. Sperling, G., & Reeves, A. (1980). Measuring the reaction time of a shift of visual attention. In R. S. Nickerson (Ed.), Attention and performance VIII. Hillsdale: Erlbaum. Steinborn, M. B., Rolke, B., Bratzke, D., & Ulrich, R. (2008). Sequential effects within a short foreperiod context: evidence for the conditioning account of temporal preparation. Acta Psychologica, 129, 297–307. Theeuwes, J. (2010). Top-down and bottom-up control of visual selection. Acta Psychologica, 135, 77–99. Thomas, E. A. (1967). Reaction-time studies: the anticipation and interaction of responses. British Journal of Mathematical and Statistical Psychology, 20, 1–29. Tipples, J. (2002). Eye gaze is not unique: automatic orienting in response to uninformative arrows. Psychonomic Bulletin & Review, 9, 314–318. Ward, R., Duncan, J., & Shapiro, K. (1996). The slow time-course of visual attention. Cognitive Psychology, 30, 79–109.

Psychological Research Wearden, J. H., & Ferrara, A. (1996). Stimulus range effects in temporal bisection by humans. The Quarterly Journal of Experimental Psychology, 49B, 24–44. Wearden, J. H., & Lejeune, H. (2008). Scalar properties in human timing: conformity and violations. The Quarterly Journal of Experimental Psychology, 61(4), 569–587.

Yantis, S. (1988). On analog movements of visual attention. Attention, Perception, & Psychophysics, 43(2), 203–206. Yantis, S. (1998). Control of visual attention. In H. Pashler (Ed.), attention (pp. 223–256). London, UK: University College Press.

123