Differential time course of implicit and explicit cueing by colour and orientation in visual search more

This article was downloaded by: [University of Birmingham, UK] On: 4 March 2011 Access details: Access Details: [subscription number 932560296] Publisher Psychology Press Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 3741 Mortimer Street, London W1T 3JH, UK Visual Cognition Publication details, including instructions for authors and subscription information: http://www.informaworld.com/smpp/title~content=t713683696 Differential time course of implicit and explicit cueing by colour and orientation in visual search Giles M. Andersona; Dietmar Heinkea; Glyn W. Humphreysa a School of Psychology, University of Birmingham, Birmingham, UK Online publication date: 08 February 2011 To cite this Article Anderson, Giles M. , Heinke, Dietmar and Humphreys, Glyn W.(2011) 'Differential time course of implicit and explicit cueing by colour and orientation in visual search', Visual Cognition, 19: 2, 258 — 288 To link to this Article: DOI: 10.1080/13506285.2010.528985 URL: http://dx.doi.org/10.1080/13506285.2010.528985 PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. VISUAL COGNITION, 2011, 19 (2), 258Á288 Differential time course of implicit and explicit cueing by colour and orientation in visual search Giles M. Anderson, Dietmar Heinke, and Glyn W. Humphreys Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 School of Psychology, University of Birmingham, Birmingham, UK Using a cued conjunction search task, Anderson, Heinke, and Humphreys (2010) demonstrated larger effects from cueing target colour than from cueing target orientation. In this study, we separated the implicit (nonexpectation-dependent) and explicit (expectation-dependent) effects of orientation and colour visual cues. In Experiment 1, we replicated the original findings for short cue durations (100Á 200 ms), demonstrating that cues matching the physical property of the target on 80% of trials exert a rapid effect on search. These early cueing effects on reaction times were supported by evidence of guidance from cues on early eye movements. Experiment 2 introduced a feature to the cue that randomly matched the colour or orientation of the target. When cue orientation was predictive, there were strong implicit effects based on whether the colour of the cue and target matched. When cue colour was predictive, there were only weak effects from the cue’s orientation. Implicit effects from cue colour remained when orientation-predictive cues were used and colour was unlikely to predict the target (Experiment 3). The data suggest that strong effects of colour cueing result from a combination of implicit and explicit processes, whereas effects from orientation cues are largely limited to the explicit guidance of visual search. Keywords: Colour; Cueing; Explicit; Implicit; Search. The process of selecting relevant visual information has been typically investigated by measuring the time taken (reaction time or RT) to find a target amongst distractor items, with the identity of the target the same across a block of trials (see Wolfe, 1998, for a review). In such tasks, participants typically learn about what they are looking for either from explicit instruction Please address all correspondence to Giles M. Anderson, School of Psychology, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK. E-mail: g.m.anderson@bham.ac.uk This work was supported by grants from the ESRC, BBSRC, EPSRC, and MRC (UK). The work was completed in partial fulfilment of a PhD by GMA. # 2010 Psychology Press, an imprint of the Taylor & Francis Group, an Informa business http://www.psypress.com/viscog DOI: 10.1080/13506285.2010.528985 IMPLICIT AND EXPLICIT CUEING IN VISUAL SEARCH 259 or during practice trials, thus constructing a template of the target. Traditionally, research emphasized the role of visual coding and segmentation of displays, with relatively little focus on how templates interact with the search process. Recently, however, researchers have investigated how search templates are set up by changing the target identity trial-on-trial while presenting information about the target prior to each search display (a precueing procedure; Anderson, Heinke, & Humphreys, 2010; Hannus, van den Berg, Bekkering, Roerdink, & Cornelissen, 2006; Moores, Laiti, & Chelazzi, 2003; Muller, Reimann, & Krummenacher, 2003; Theeuwes, ¨ Reimann, & Mortier, 2006; Vickery, King, & Jian, 2005; Wolfe, Horowitz, Kenner, Hyle, & Vasan, 2004). Both verbal and visual cues have been shown to affect search behaviour. Wolfe et al. (2004) compared RTs in a conjunctive search when (1) a visual representation of the target (e.g., a red vertical stimulus) was presented prior to each trial and (2) verbal cues preceded each search trial (e.g., the words ‘‘RED VERTICAL’’). Search was facilitated following either type of cue compared to when no information about the target was available, indicating that both physical and abstract information (from verbal cues) can be used to set a template of the search target. However, there was larger facilitation from visual cues, with this effect reaching a maximum when there was a short stimulusÁonset asynchrony (SOA) between the cue and the search display (200 ms) and then decreasing at longer SOAs. Guidance from verbal cues took longer and still had an increasing influence at longer SOAs (e.g., 800 ms). Wolfe et al. suggested the advantage of visual representations was due to ‘‘implicit top-down guidance’’ (i.e., without an explicit expectation being generated from the cue), with the physical nature of the cue enhancing the guidance of attention to the target. Vickery et al. (2005) also demonstrated an advantage for visual versus verbal cues. Using separate searches based on abstract shapes and real-world items, the authors found the efficacy of the cue was dependent on its similarity to the target image. RTs were shortest when the cueing image exactly matched the target; this benefit reduced when the cue differed in size or orientation, and decreased further when the cue was presented in word form (e.g., verbal cues). Vickery et al. suggested the added facilitation from exact visual cues was a consequence of the efficient creation of the target template being reliant on a detailed visual representation being held in working memory. The relative facilitation from exact versus nonexact visual cues, however, decreased following a longer cue leading time (1000 ms), suggesting that the visual details of the cue lost their influence over time. In earlier more recent study, we (Anderson et al., 2010) compared the effects of visual and verbal cues on a difficult conjunction search, in which participants searched for one of two possible targets. Participants were given cues that predicted the colour or the orientation of the target (valid on 80% of the trials). RTs were shorter when the critical feature of a visual cue Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 260 ANDERSON, HEINKE, HUMPHREYS matched that of the target, whereas search was slower when the cue and target differed in the cued dimension. Comparable effects were found with verbal rather than visual cues and in both cases stronger modulation was evident from colour rather than orientation cues. When visual cues are used in search, however, it is often unclear whether effects may result from implicit (i.e., nonexpectation-based) properties of the cues, which happen to match the target on a valid cue trial, rather than from explicit, conscious expectancies. Prior work has provided evidence for both explicit (prediction-based) and implicit (nonprediction-based) effects on search from visual features of a stimulus in working memory. Soto, Heinke, Humphreys, and Blanco (2005) used an orientation-defined search, with each of the search items contained within a search-irrelevant coloured stimuli. When a pretrial cue was held in working memory, search RTs were affected when the colour or shape of the cue matched a feature of a stimulus containing a search item, even when the memorized feature never matched that of the stimuli containing the target. Soto and Humphreys (2009) also showed that, when the shape of an item was maintained in working memory, search was affected by the item’s colour (for contrasting results, see Olivers, Meijer, & Theeuwes, 2006). They proposed that the whole cue was coded so that both features (colour and shape) entered working memory and influenced search, irrespective of whether or not memory for colour was required. It should be noted, however, that Soto et al. found no effect on search of merely attending to the cue, suggesting these effects were reliant on maintaining the cue in working memory. Robust implicit effects of the colour of the stimuli have also been evident with no overt requirement to memorize the features of the cue. Kristjansson ´ (2006; see also Goolsby & Suzuki, 2001; Huang, Holcombe, & Pashler, 2004; Kumada, 2001; Maljkovic & Nakayama, 1994; Meeter & Olivers, 2006; Muller, Krummenacher, & Heller, 2004; Theeuwes et al., 2006) found RTs ¨ were shorter in a feature-singleton search task when the colour, orientation, or spatial frequency of the target was repeated trial-on-trial, compared to when the target-feature changed. This facilitation was strongest when the target colour was repeated. Corresponding intertrial effects have also been found on conjunction search tasks when the exact target is repeated on consecutive trials (e.g., Becker & Horstmann, 2009; Kristjansson, 2006; ´ Kristjansson, Wang, & Nakayama, 2002). Whether intertrial carryover ´ effects reflect only implicit or also explicit (expectancy-based) effects remains controversial (contrast Maljkovic & Nakayama, 1994, with Muller et al., ¨ 2003). In the present research we sought to tease apart implicit and explicit effects on search from different cues, focusing on whether the previously observed stronger effects from colour over orientation cues (Anderson et al., 2010) reflected contributions that occur implicitly, from the mere encoding of the Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 IMPLICIT AND EXPLICIT CUEING IN VISUAL SEARCH 261 Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 visual properties of (visual) cues. We used orientation- and colour-defined targets that were matched for bottom-up saliency (measured in single feature search conditions; cf. Bacon & Egeth, 1997; see Anderson et al., 2010, for details). Baseline experiments were undertaken in which we examined the time course of effects from visual cues (Experiments 1a and 1b), establishing whether cues affect the guidance of eye movements to the target (Experiment 1c). A nonpredictive feature was added to the cue, enabling us to measure if this also modulated search (implicitly). Experiment 3 subsequently investigated whether a nonpredictive feature of the cue affected search when the likelihood of the feature matching the target was below chance. EXPERIMENT 1: EXPLICIT CUEING FOLLOWING VARYING CUE DURATIONS Previous research has shown larger modulation of visual search following cues predicting the colour of the target compared to when information about its orientation was cued (Anderson et al., 2010; Muller et al., 2003). ¨ However, the predictive information available following visual cues (as used in Anderson et al.) can be viewed as both explicitly represented (e.g., using expectations from a template of the target to guide search; Duncan & Humphreys, 1989) and implicitly represented (following activation of a physical property of the cue, even if that property is not used to predict the target). Implicit priming from the physical properties of the cue may influence search in a bottom-up manner, from mere activation of the perceptual system. Guidance from this priming may act more rapidly than top-down, expectancy-based guidance (e.g., Wolfe et al., 2004). Hence it was important to map out the time course over which a cue could influence search, before implicit and explicit effects of the cue were teased apart. This was the aim of Experiment 1, where the duration of an explicit visual cue to the colour or orientation information of the target was varied. On each trial of the search task, either of two targets could appear*blue horizontal or green vertical bars; there were always two types of distractor (blue vertical and green horizontal, with a 50:50 ratio for each distractor type). A pretrial stimulus cue predicted either the colour (a green or blue patch) or orientation (horizontal or vertical white line) of the target at 80% validity (cf. Mu ¨ller et al., 2003). The presentation duration of the cue was varied between 100 ms and 200 ms (Experiment 1a) and between 200 ms and 1200 ms (Experiment 1b). Stimulus-driven dimensional differences should be evident at shorter durations, whereas variation following longer cues may reflect top-down guidance from the cue. 262 ANDERSON, HEINKE, HUMPHREYS Experiment 1a Method Participants. Eighteen University of Birmingham students, five male, 13 female, aged 18Á28 (average 21.95) took part. All had self-reported normal or corrected-to-normal vision; had normal colour perception, assessed using Ishihara’s (1981) Tests for Colour-Deficiency; and were naıve ¨ to the purpose of the experiment. Design. This was a within-participant design, with four independent variables: Cue validity (valid, invalid), cue duration (100 ms, 200 ms), cued dimension (colour, orientation), and target type (blue horizontal, green vertical). A compound search task was used. The dependent variable was the RT and accuracy for responses to a symbol which appeared on the target (see Olivers & Meeter, 2006). Stimuli. The pretrial fixation stimulus was a white O, 0.6cm in diameter (visual angle of 0.578), line width of 0.1cm (0.0958). The stimulus presented prior to each cued trial was one of four stimuli: A blue patch, a green patch, a white horizontal bar, or a white vertical bar. The patches were colour cues in the shape of filled circles, all with the diameter of 0.35 cm (0.348) coloured to match the stimuli in the search array (green or blue). The physical dimensions of the orientation cues (vertical or horizontal) were also the same as those for the search stimuli, although the cue stimuli were white. All items were presented on a black background, with the array comprising blue vertical or green horizontal bars as distractors and either a blue horizontal or green vertical bar as the target. These bars were 0.8 cm (0.778) long)0.2 cm (0.198) wide, with symbols positioned centrally. To minimize interactions with stimuli orientation, the symbols were symmetrical and either ‘‘x’’s or ‘‘'’’s. The symbols measured 0.2 cm (0.198))0.2 cm (0.198))0.025 cm wide (0.0248); they were distributed across all stimuli so each symbol was added to half the distractors. The symbol present on the target varied across each trial, with equal likelihood. Exploratory experiments were undertaken to ensure that targetÁdistractor saliency was matched across the colour and orientation dimensions (see Anderson et al., 2010). The CIE (International Commission on Illumination) colour and luminance values of the stimuli were measured by a SpectraCAL spectroradiometer (Cambridge Research Systems) and are listed in Table 1. Apparatus. The stimuli were presented at 1024)768 resolution on a 17-inch colour Samsung SyncMaster 793s monitor, driven by an Intel Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 IMPLICIT AND EXPLICIT CUEING IN VISUAL SEARCH 263 TABLE 1 The CIE colour values (x,y) as well as the luminance of the stimuli in Experiments 1Á3 as measured by a SpectraCAL spectroradiometer (Cambridge Research Systems) Colour All experiments (except 1c) Blue Green White Black Grey Experiment 1c Blue Green White Black Grey x y Luminance (cdmÁ2) 0.2495 0.2839 0.2815 0.302 0.3803 0.2878 0.3868 0.3099 0.3731 0.3124 47.93 52.82 157.9 8.635 112.7 Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 0.2387 0.3026 0.2829 0.3102 0.2845 0.2465 0.4362 0.2963 0.3626 0.2972 8.145 10.08 77.04 3.525 42.18 Pentium 4 PC with a Radeon 9000 AGP Pro video card. Displays were generated by an E-Prime program (Schneider, Eschman, & Zuccolotto, 2002) that recorded RTs and accuracy via a standard UK keyboard. Audio feedback was provided by stereo Cambridge Soundworks speakers. Participants sat approximately 0.6 m from the screen in a well-lit room. Procedure. Participants undertook 15 practice trials, during which performance was not recorded, followed by two blocks of cued trials (a total of 240 trials). Initially, a fixation circle was present for 1000 ms. This was followed by a cue stimulus that was displayed for either 100 ms or 200 ms, then a 100 ms interstimulus interval (ISI) before an array of stimuli was presented consisting of one target and 14 distractors. The items were presented randomly within an invisible circle of diameter 5.5 cm (2.628) with 21 possible positions. The circle was in the middle of the screen and the positions of the stimuli were staggered ('/(0.18) horizontally and vertically to minimise interactions between distractors. Half the distractors were blue vertical bars and the other half were green horizontal bars. Due to high error rates during piloting, participants were informed of the nature of the targets prior to the experiment (green vertical or blue horizontal bars), with graphical reminders presented adjacent to the computer monitor during the experiment. Participants indicated the feature present on the target (‘‘'’’ or ‘‘x’’) by pressing either ‘‘Z’’ or ‘‘M’’ on the computer keyboard (the key assignment was reversed for half the participants). Feedback was provided. If the response was correct, participants heard a medium pitched sound and the word ‘‘Correct’’ was 264 ANDERSON, HEINKE, HUMPHREYS displayed. If incorrect, a lower note was played and the word ‘‘Incorrect’’ was displayed. The time until participants’ response (RT) and the accuracy of the response was recorded. On half the trials the cue predicted the colour of the target (colour cues 80% valid), on half it predicted the target’s orientation (orientation cues 80% valid; Figure 1). A green or blue patch was used as a colour cue, and a white vertical or horizontal line formed the orientation cue with the nature of the cue randomized trial-on-trial. As well as the possible target configuration, participants were informed prior to the cued blocks that the majority of the cues were valid. Trials with the same presentation time were blocked, with block order counterbalanced across participants. Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 Results Due to the difficulty of the task, several participants required breaks in the middle of a block of trials, and on those occasions a trial was left to run until the break concluded. For all analyses, these trials were removed (RTs5000 ms). Less than 1% of data were removed as a consequence. To control the familywise error rate, Huynh-Feldt adjustments were used where necessary. All pairwise comparisons included Bonferroni adjustments and were measured as significant at the pB.05 level. RTs. To maximize data following invalid cues, trials were pooled across target type (blue horizontal and green vertical) and medians for each participant were calculated. The data were separated into whether the colour or orientation of the target was cued and whether this information was valid or invalid. Group means are shown in Figure 2. Colour cues Orientation cues 1000ms Varied 100ms ISI Until response Figure 1. Simplified trial timelines for Experiment 1. Blue stimuli are shown as black, green as grey, white as outline, black background, and light grey symbols as white. A green vertical target is circled, although it is equally likely that a blue horizontal target could be present instead. IMPLICIT AND EXPLICIT CUEING IN VISUAL SEARCH 265 3500 Exp. 1a: valid colour Exp. 1a: invalid colour 3000 Exp. 1a: valid orientation Exp. 1a: invalid orientation RTs (ms) 2500 Exp. 1b: valid colour Exp. 1b: invalid colour Exp. 1b: valid orientation 2000 1500 Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 Exp. 1b: invalid orientation 1000 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 Cue duration (ms) Figure 2. Mean RTs ('/( standard error) from Experiments 1a and 1b, separated by cue duration, cue dimension, and cue validity. Data from Experiment 1a are shown in grey; Experiment 1b data are shown in black. A three-factor ANOVA (cue duration, cue dimension, cue validity) found a main effect of validity, F(1, 17)0103.819, MSE0475212.39, pB.001. RTs on valid trials (1776 ms) were shorter than those on invalid trials (2947 ms). There was a Cue duration)Cue validity interaction, F(1, 17)04.999, MSE0 88000.14, p0.039, with a larger difference between RTs on valid trials and invalid trials (a ‘‘validity effect’’) when cues were presented for longer (200 ms; a ‘‘validity effect’’ of 1281 ms, pB.001) compared to when cues were presented for a shorter duration (100 ms; an effect of 1060 ms, pB.001). A Cue dimension)Cue validity interaction was also evident, F(1, 17)016.774, MSE095495.30, p0.001. There was a larger validity effect when the colour of the target was cued (an effect of 1382 ms, pB.001) compared to when orientation was cued (an effect of 960 ms, pB.001). The three-way interaction did not reach significance, F(1, 17)01.453, MSE059152.09, p0.245, indicating that the larger modulation following colour relative to orientation cues did not vary with duration. Separating the data by validity, analysis of valid RTs showed a main effect of cue dimension, F(1, 17)042.842, MSE043502.66, pB.001. RTs were shorter following colour cues (1615 ms) compared to when target orientation was cued (1937 ms), with this advantage for cue colour unaffected by cue duration, Cue duration)Cue dimension interaction (FB1). The data from invalid trials showed no main effects or interactions (all ps.2). 266 ANDERSON, HEINKE, HUMPHREYS Accuracy. Error rates followed the pattern of RTs with no speedÁ accuracy tradeoff evident. Mean accuracy is presented in Table 2. Experiment 1b Method The methodology for Experiment 1b matched that of Experiment 1a except that the cue duration was varied from 200 ms to 1200 ms. Specific differences are outlined later. Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 TABLE 2 Mean accuracy (%) for all experiments Cue duration (ms) Experiment 1a Cue type Valid colour Invalid colour Valid orientation Invalid orientation Valid colour Invalid colour Valid orientation Invalid orientation Valid colour Invalid colour Valid orientation Invalid orientation Valid colour, orientation matches Valid colour, orientation differs Invalid colour, orientation matches Invalid colour, orientation differs Valid orientation, colour matches Valid orientation, colour differs Invalid orientation, colour matches Invalid orientation, colour differs Valid orientation, colour matches Valid orientation, colour differs Invalid orientation, colour matches Invalid orientation, colour mismatch 100 95 95 96 95 200 95 93 96 95 95 97 96 97 98 99 95 97 98 96 96 97 98 97 97 96 1200 1b 97 93 95 95 1c 2 3 97 97 98 98 97 97 97 96 96 97 98 97 97 96 98 96 Array size Search Type Appendix Colour-defined Orientation-defined Conjunction 5 * * 91 9 95 95 93 IMPLICIT AND EXPLICIT CUEING IN VISUAL SEARCH 267 Participants. Eighteen University of Birmingham students, five male, 13 female, aged 18Á25 (average 19.83) took part in the experiment. Procedure. Cues were presented for either 200 ms or 1200 ms. Results RTs. The data were cleaned and median RTs for valid and invalid trials were calculated as in Experiment 1a. Group means are shown in Figure 2. A three-factor ANOVA found main effects of cue dimension, F(1, 17)05.043, MSE061561.67, p0.038, and cue validity, F(1, 17)0157.345, MSE0 354095.19, pB.001. Two two-way interactions were evident: Cue duration)Cue dimension, F(1, 17)04.758, MSE060143.63, p0.044, and Cue dimension)Cue validity, F(1, 17)010.828, MSE0247583 p0.004, as well as a three-way interaction, F(1, 17)06.307, MSE072014.32, p0.022. To unpack the three-way interaction, the RTs on valid trials were analysed separately to RTs following invalid cues. Valid trial data showed only a main effect of cue dimension, F(1, 17)021.256, MSE0113279.42, pB.001; RTs were shorter following valid colour cues (1622 ms) compared to valid orientation cues (1988 ms). Data from invalid trials revealed a Cue duration)Cue dimension interaction, F(1, 17)06.102, MSE0119740.95, p0.024, with longer RTs following invalid colour cues compared to invalid orientation cues at the shorter cue duration (200 ms duration; a difference of 382 ms, p0.018) but no corresponding difference when the cue was presented for longer (1200 ms; a difference of 21 ms, p0.858). No other main effects or interactions reached significance (all FsB1). Accuracy. Error rates followed the pattern of RTs with no evidence of a speedÁaccuracy tradeoff. Mean accuracy is presented in Table 2. Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 Discussion As in Anderson et al. (2010), there was a strong modulation of attention when both the physical and the predictive properties of a visual cue either matched (valid cues) or differed from (invalid cues) the following search target. Valid cues facilitated search compared to invalid cues. These effects were stronger from colour than from orientation cues. RTs were shorter when the colour of the cue matched that of the target compared to when the cue and target possessed the same orientation, though differences between colour and orientation cues were reduced when cues were invalid. RTs were longer following invalid colour cues compared to invalid orientation cues when the cue was presented for 200 ms (cf. Anderson et al., 2010), but this effect was diminished at shorter and longer cue durations (e.g., 100 ms and 1200 ms, respectively). Shorter RTs following 268 ANDERSON, HEINKE, HUMPHREYS valid colour cues, however, were evident at all cue durations. The data point to an early effect of colour cueing, which is equated over time by the effects of orientation cues. Experiment 1c The effects of the cues on RTs, established in Experiments 1a and 1b, could arise for various reasons. The cue could guide selection of the target. Alternatively, there may be no gain in search efficiency; rather each item may be selected serially but the cue may remind participants of the target templates. To tie down whether the cue facilitates target selection, in Experiment 1c we assessed its influence on early eye movements as participants searched displays. If the cue only affects the confirmation of each selected item as being a target or not, we would expect no effect of the cue on eye movements. In contrast, if there was guidance of search, the cue should modulate the frequency of eye movements to targets. Moreover, if there were effects on the initial guidance of selection, the cue should influence first eye movements. Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 Method Experiment 1c matched Experiments 1a and 1b when cues were presented for 200 ms, except as outlined later. Participants. Nineteen University of Birmingham students, one male, 18 female, aged 18Á21 (average 18.89) took part. Design. Position of eye gaze was measured, alongside RTs and accuracy. Apparatus. The stimuli were presented on a 22-inch colour CRT monitor (ViewSonic P225f, 2004) at 1024)748 pixel resolution and a refresh rate of 100 Hz. Participants used a chinrest 0.6 m from the screen in a dimly lit room with windows blacked out to avoid luminance changes. Chinrest and monitor height was adjusted for each participant so gaze was central to the display screen. Eye movements were recorded using an SMI infrared Remote Eyetracking Device III (SMI RedIII; SensoMotoric Instruments GmbH, Germany 2002Á2004). The gaze position accuracy was 0.58 visual angle and sampling rate 50 Hz (every 20 ms). The eyetracking camera was linked to a computer running IViewX (version 1.07.00) software that calibrated the camera and collected eye movement data. IViewX was synchronized with E-Prime software on the display computer via an Ethernet cable. IMPLICIT AND EXPLICIT CUEING IN VISUAL SEARCH 269 Stimuli. All stimuli were presented on a grey background to aid eyetracking. The fixation circle was 0.6 cm diameter (visual angle of 0.578 at 0.6 m viewing distance). The nature of possible targets and distractors were as previously. Prior to the experiment, we ran pilot studies to equate search efficiency for colour- and orientation-defined targets against single feature baseline conditions (see Appendix for details). The resultant colour values are shown in Table 1. Individual stimuli were 1 cm (18) long)0.3 cm (0.38) wide. As previously, the colour coordinates for colour cues (coloured circles with diameters of 0.8 cm, 0.88) and the dimensions of the orientation cues matched those of the search items. Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 Procedure. Trials matched those in Experiments 1a and 1b in which cues were presented for 200 ms. However, the visual reminders of the two possible targets adjacent to the computer monitor were only available on practice trials during which eye movements and behavioural data were not recorded. To increase the accuracy of the eyetracking, only nine stimuli were included in the search array: Either a blue horizontal or green vertical target, with four green vertical and four green horizontal bar distractors. There were 30 to 40 practice trials, followed by two blocks of 120 trials. Results RTs. The data were cleaned as previously and mean of median RTs are shown in Figure 3. A two-factor ANOVA (cue dimension, cue validity) 2800 Colour cues 2600 2400 2200 RTs (ms) 2000 1800 1600 1400 1200 1000 Valid Cue validity Invalid Orientation cues Figure 3. Mean RTs ('/( one standard error) from Experiment 1c, separated by the validity of the cue and the cue dimension (colour or orientation). As all cues were presented for 200 ms, the data were not separated by this factor. 270 ANDERSON, HEINKE, HUMPHREYS showed a main effect of cue dimension, F(1, 18)014.871, MSE052999.36, p0.001, and cue validity, F(1, 18)082.576, 144560.877, pB.001. RTs were shorter following colour compared to orientation cues (1983 ms vs. 2193 ms). Search was also facilitated on valid compared to invalid trials (1680 ms vs. 2497 ms). Figure 3 suggests a larger validity effect from colour compared to orientation cueing; however, the interaction did not reach significance, F(1, 18)01.473, MSE076606.897, p0.242. Accuracy. There was no speedÁaccuracy tradeoff. The mean accuracy data are shown in Table 2. Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 Eye movements. For each trial, eye movements were recorded from the onset of the search array until response. A fixation was classified when the speed of the eye movement remained below 50 visual degrees per second for 100 ms, with pilot data indicating these parameters as optimum for the search display. Data recorded during eye-blinks and off-screen eye movements were discarded, as were fixations detected within 80 ms of array onset (see van Zoest, Donk, & Theeuwes, 2004). Following this, for each fixation we calculated the identity of the nearest stimulus (cf. Williams & Reingold, 2001). As with the RT data, we removed inaccurate trials and those with RTs of more than 5000 ms. The number of fixations per trial varied depending on search efficiency. At least 80% of the trials from each participant contained two fixations or more so that only these fixations could be analysed. Trials with fewer fixations were discarded. One participant was removed due to calibration error. Frequency of fixating the target. The mean frequency of target fixations was calculated by summing the number of trials on which a target was fixated and then dividing this by the total number of trials. The data were pooled across target type, as in the RT analysis, but were separated by fixation number as well as cue validity. The data were then adjusted for chance. The probability of a random fixation being directed to a target was subtracted from the relevant frequencies at both fixations. This value was 1/9 (there were nine search items) and was the same for Fixation 1 and Fixation 2. Data were pooled across target type, as in the RT analysis, and mean frequencies across participants are shown in Figure 4. A three-factor ANOVA (fixation number, cue dimension, cue validity) found a significant effect of cue validity, F(1, 16)023.307, MSE00.049, pB.001, with higher frequency of target-fixations on valid trials (an adjusted frequency of 0.291) compared to invalid trials (an adjusted frequency of Á0.077), and fixation number, F(1, 16)0139.914, MSE0 0.015, pB.001. This ‘‘validity effect’’ increased across fixations, Fixation number)Cue validity interaction, F(1, 16)05.702, MSE00.023, p0.03. IMPLICIT AND EXPLICIT CUEING IN VISUAL SEARCH Valid colour Invalid colour 0.5 0.4 0.3 0.2 0.1 0 -0.1 Fixation 1 Fixation 2 271 0.6 Adjusted frequency of target-fixations Valid orientation Invalid orientation Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 Figure 4. Mean frequency of a fixation landing on the target ('/( one standard error) from Experiment 1c, separated by fixation number, validity of the cue (valid or invalid), and the cue dimension (colour or orientation). The data were adjusted so a score of zero equals chance frequency. There was a validity effect of 0.121 at first fixation (p0.003), and at Fixation 2 the effect increased to 0.246 (pB.001). There was also an interaction between cue dimension and cue validity, F(1, 16)05.253, MSE00.016, p0 .036. There was a larger effect of cue validity from colour (an effect of 0.234, pB.001) compared to orientation cues (an effect of 0.134, p0.004); however, this did not vary between Fixations 1 and 2 (three-way interaction, FB1). Discussion Although there were similar effects from colour and orientation cues (in contrast to Experiments 1a and 1b; Anderson et al., 2010), an advantage following colour cueing was evident in the eye movement data. Although both valid colour and valid orientation trials increased fixations to the target compared to invalid trials, the validity effect was stronger for colour cues. The eye movement data suggest that cues do guide search. Robust effects on target fixations occurred as early as Fixation 1, increasing at subsequent fixations. Displays contained nine items and, on a serial search account, there should be on average 4.5 fixations before a target is selected. That the cues influenced earlier fixations provides strong evidence that cues affected how search was guided to the target, rather than just acting as reminders of target identity. A more pronounced effect of colour compared with orientation cues was apparent on eye movements as well as RTs (Experiments 1a and 1b). Experiments 2 and 3 set out to test whether the advantage for colour cues 272 ANDERSON, HEINKE, HUMPHREYS arises from a stronger explicit cueing effect (stronger effects of top-down prediction) or whether there is an implicit (nonpredictive) component, based on the perceptual encoding of the cue’s colour. EXPERIMENT 2: NONPREDICTIVE CUE FEATURES In Experiment 2, we introduced cues where both a predictive feature and a nonpredictive feature were present with participants asked to ignore the nonpredictive feature. The cues were coloured lines. In one session the cue colour was predictive (80% valid), and its orientation matched1 the target 50% of the time. In a separate session, cue orientation was predictive (80% valid) and its colour matched the target at 50%. Effects of the nonpredictive feature may provide evidence for implicit effects from the physical nature of the cue (bottom-up implicit processes). Cue duration was again varied to assess possible differences in the time course of any implicit or explicit effects of the cue. Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 Method Experiment 2 matched Experiment 1a except as outlined later. Participants. Eighteen University of Birmingham students, three male, 15 female, aged 18Á29 (average 22.2) took part. Design. There were five independent variables: Dimension of predictive feature of the cue (colour, orientation), validity of predictive feature of the cue (valid, invalid), nonpredictive feature status (whether it matches or differs from the target), cue duration (100 ms, 200 ms), and target type (blue horizontal, green vertical). Stimuli. The cueing stimuli were blue horizontal, green vertical, blue vertical, or green horizontal bars. The colour and spatial dimensions of the cues matched those of the stimuli used in the search array, but without the response-defining symbols. Procedure. Participants took part in two sessions, a minimum of 24 hours apart. In each session, participants undertook two blocks of 120 cued trials, with cues presented for 100 ms in one block and 200 ms in the other. Block order was counterbalanced across participants. Before the We differentiate between cue validity, where there is a predictive link between cue feature and the target, and whether the cue matches or differs to the target where any link was nonpredictive. 1 IMPLICIT AND EXPLICIT CUEING IN VISUAL SEARCH 273 Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 first block there were 15 practice trials, where RTs and accuracy were not recorded. The cueing stimulus on each trial could be a blue horizontal, green vertical, green horizontal, or blue vertical bar. In one experimental session, the colour of the cue was predictive so that it was the same as the target 80% of the time and differed 20% of the time (cf. Muller et al., 2003), and the ¨ orientation of the cue (the nonpredictive feature) randomly matched the target (50%). The opposite was true in the other session (orientation predictive at 80%/20%, cue colour nonpredictive) with session order counterbalanced across participants. The predictive feature of the cue could be either valid or invalid while concurrently the nonpredictive feature could match or differ from that of the target (50% match, 50% mismatch). For each session, participants were informed of the predictive nature of the relevant cue feature and asked to ignore the nonpredictive feature. Results RTs. The data were cleaned as in Experiment 1. As previously, trials were pooled across target type and medians for each participant in each condition were calculated. Group means are shown in Figure 5. A four-factor ANOVA (cue duration, predictive cue dimension, predictive feature validity, nonpredictive feature status) found a main effect of the validity of the predictive feature, F(1, 17)0181.802, MSE0530660.432, pB.001. RTs when the predictive property of the cue was valid (1459 ms) were 3500 Colour valid (80%), orientation matches (50%) Colour valid (80%), orientation differs (50%) 3000 Colour invalid (20%), orientation matches (50%) Colour invalid (20%), orientation differs (50%) Orientation valid (80%), colour matches (50%). RTS (ms) 2500 2000 Orientation valid (80%), colour differs (50%) Orientation invalid (20%), colour matches (50%) Orientation invalid (20%), colour differs (50%) 1500 1000 0 50 100 150 200 250 300 350 400 Cue duration (ms) Figure 5. Mean RTs ('/( one standard error) from Experiment 2, separated by predictive dimension, validity of predictive cue feature (valid or invalid), nonpredictive feature status (whether it matches or differs from the target), and cue duration. 274 ANDERSON, HEINKE, HUMPHREYS shorter compared when the cue was invalid (2616 ms). There was also a main effect of the status of the nonpredictive feature, F(1, 17)020.998, MSE0 170205.42, pB.001, with search facilitated when the feature matched the target compared to when the two differed (1926 ms vs. 2149 ms). There were several significant interactions. The Cue dimension)Validity of predictive feature interaction, F(1, 17)013.754, MSE0267910.24, p0.002, was due to a larger validity effect when colour was the predictive feature of the cue (difference between valid and invalid RTs of 1384 ms, pB.001) compared to when the orientation of the cue was predictive (a difference of 931 ms, pB .001). The other interactions were: Predictive cue dimension)Nonpredictive feature status, F(1, 17)06.938, MSE0211716.82, p0.017, Predictive cue validity)Nonpredictive feature status, F(1, 17)07.29, MSE098474.8, p0.015; and Predictive cue dimension)Predictive feature validity)Nonpredictive feature status, F(1, 17)05.851, MSE047918.97, p0.027. The data were decomposed by predictive cue dimension. The colourpredictive data reflected a robust main effect of predictive cue validity, F(1, 17)0326.985, MSE0210819.66, pB.001, with shorter RTs on valid trials (1321 ms) compared to invalid trials (2705 ms). There was also a borderline significant effect of the status of the nonpredictive feature, e.g., orientation, F(1, 17)04.099, MSE056310.85, p0.059, with search facilitated when cue orientation matched the target compared to when there was a mismatch (1973 ms vs. 2053 ms). No other main effects or interactions reached significance (all ps.2). Analysis of orientation-predictive trials found robust main effects of both predictive feature validity, F(1, 17)053.127, MSE0 587751.02, p 0 001, and the status of the nonpredictive feature (colour), F(1, 17)014.871, MSE0325611.4, p0.001. There was also a Predictive feature validity)Nonpredictive feature status interaction, F(1, 17)010.459, MSE090615.91, p0.005. When the predictive feature (orientation) was valid, RTs were shorter when the nonpredictive feature (colour) of the cue also matched that of the target compared to when it differed (a difference of 528 ms, pB.001). When the orientation of the cue was invalid, however, there was a sizeable effect of the nonpredictive cue colour but this did not reach significance (a difference of 203 ms, p0.131). To directly compare the effects when the cue’s colour and when its orientation were predictive, RTs on valid trials and invalid trials were analysed separately. The analysis of RTs following valid cues showed main effects of the predictive cue dimension, F(1, 17)021.409, MSE0127357.71, pB.001, and status of the nonpredictive feature, F(1, 17)042.134, MSE088948.69, pB .001. There was also a Predictive cue dimension)Nonpredictive feature status interaction, F(1, 17)021.807, MSE069536.72, pB.001. RTs were shorter when cue colour was valid and its (nonpredictive) orientation did not match the target’s (e.g., a green horizontal cue followed by a green vertical target) compared to when its orientation was valid and cue colour differed from the Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 IMPLICIT AND EXPLICIT CUEING IN VISUAL SEARCH 275 target, e.g., a blue vertical cue followed by a green vertical target (a difference of 480 ms, pB.001). However, there was no difference between the effects of valid colour- and orientation-predictive cues when the nonpredictive features also matched the target (a difference of 70 ms, p0.344). Analysis of RTs from invalid trials showed a borderline main effect of nonpredictive feature status, F(1, 17)03.028, MSE0179731.527, p0.1, reflecting a trend towards longer RTs when the feature did not match the target (2678 ms) compared to when the two matched (2555 ms). No other significant main effects or interactions reached significance (all ps.1). Accuracy. No speedÁaccuracy tradeoff was evident. The mean accuracy data are presented in Table 2. Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 Discussion The results demonstrate robust differences between the implicit effects of the colour and orientation of the cue, effects which did not vary across cue durations. When the orientation of the cue was predictive, the cueing effect was influenced by whether the nonpredictive colour matched the target or not. Search was facilitated when the cue’s colour was the same as that of the target, compared to when the cue and target differed in colour. This ‘‘matching effect’’ was larger when orientation was valid (and effect of 528 ms) compared to when it was invalid (an effect of 200 ms). In contrast, the effect of nonpredictive orientation was smaller (an effect of 80 ms averaged across valid and invalid trials). The findings suggest that the colour of the cue was processed early as there was no variation with cue duration and these processes occurred regardless of whether there was any predictive benefit, indicating implicit mechanisms may be involved (e.g., Soto & Humphreys, 2009). Indeed, when both cue features matched the target there were similar effects whether colour or orientation was the predictive dimension, suggesting that the colour of the cue has a similar effect regardless of whether its relationship with the search target was explicit or implicit. Although robust bottom-up implicit effects were evident, particularly from the colour of the cue, we cannot conclude that these effects were purely stimulus driven. For this to be the case, we would expect participants to be unable to inhibit their influence on RTs (cf. Posner, Cohen, & Rafal, 1982). In the present case, participants could still be attending to the colour information, even though instructed not to do so. The colour matched that of the target on 50% of the orientation-cue trials and there was therefore little disincentive from using the colour as well as the orientation of the cue. To provide a stronger test of automatic effects, in Experiment 3 we tested bottom-up implicit effects of colour cues while including a cost to attending to this feature (cf. Kean & Lambert, 2003). The orientation of the cue was 276 ANDERSON, HEINKE, HUMPHREYS Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 predictive at 80% as previously (see Experiments 1 and 2); however, the colour of the cue was contrapredictive, only matching that of the target on 20% of trials. Attending to cue colour would therefore be detrimental to performance (e.g., Soto et al., 2005). Cues were either short (100 ms) or long (1200 ms). We would expect automatic effects of cue colour to be evident at shorter cue durations, with search reflexively guided to stimuli sharing the colour of the cue. However, any explicit processing of the colour of the cue should be apparent following longer cues (cf. Posner et al., 1982). If the contrapredictive feature is processed in a top-down manner using it to guide search, a longer time between cue presentation and search may allow participants to override the effects of cue colour once they became aware of the cost associated with the cue’s colour (cf. Neely, 1997). Explicit processing of the cue colour would be demonstrated by either a reduction in its effects on search or a switching of the effect of the cue’s colour (with participants searching for a target matching the opposite colour to that of the cue). In the latter case, there would be a benefit when the colour of the cue did not match the target, compared to when it did (cf. Posner et al., 1982). EXPERIMENT 3: IMPLICIT CONTRAPREDICTIVE CUE COLOUR Method Experiment 3 replicated Experiment 2 except for the details outlined later. Participants. Twenty-one female University of Birmingham students aged 18Á37 (average 21.95) took part. One participant was removed due to accuracy of less than 90%. Procedure. The trials were the same as those in Experiment 2 when cue orientation was predictive, with certain alterations. Cue duration was either 100 ms or 1200 ms and the probability of the colour of the cue matching the target was reduced from 50% to 20%. Therefore, while the predictive information from the orientation of the cue was 80% correct, the cue colour was only the same as that of the target on 20% of trials. Participants undertook four blocks of 100 experimental trials, with the initial block preceded by 15 practice trials. On two blocks the cue duration was 100 ms; the cue was presented for 1200 ms on trials during the other two blocks. Block order was counterbalanced across participants. Results RTs. The data were cleaned as in Experiment 2 and median RTs were calculated as previously. Group means are shown in Figure 6. IMPLICIT AND EXPLICIT CUEING IN VISUAL SEARCH 3500 277 Orientation valid (80%), colour matches (20%) 3000 Orientation valid (80%), colour differs (80%) RTs (ms) 2500 Orientation invalid (20%), colour matches (20%) 2000 Orientation invalid (20%), colour differs (80%) 1500 Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 1000 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 Cue duration (ms) Figure 6. Mean RTs ('/( one standard error) from Experiment 3, separated by the validity of the predictive cue feature (orientation), contrapredictive colour status (whether it matches or differs with the target), and cue duration. A three-factor ANOVA (cue duration, orientation validity, status of contrapredictive cue colour) found a main effect of orientation validity, F(1, 19)056.908, MSE0793939, pB.001; RTs where shorter on valid trials compared to invalid trials (1814 ms vs. 2877 ms). There was an interaction between orientation validity and cue duration, F(1, 19)06.402, MSE0 102209.12, p0.02, a borderline significant interaction between orientation validity and the status of the contrapredictive cue colour, F(1, 19)04.115, MSE055904.95, p0.057, and a robust three-way interaction, F(1, 19)0 5.569, MSE055199.78, p0.029. To unpack the three-way interaction, data were split by the validity of the predictive feature (orientation). Analysis of RTs following valid cues showed a main effects of cue duration, F(1, 19)04.823, MSE071934.24, p0.041, and the status of contrapredictive cue colour, F(1, 19)05.9, MSE053423.05, p0.025. RTs were reduced following longer cues compared with shorter cues (1200 ms cues vs. 100 ms cues: RTs, 1748 ms vs. 1800 ms). Search was also facilitated when the colour of the cue matched the target compared to when it did not (1752 ms vs. 1877 ms). There was a borderline significant interaction between cue duration and the status of contrapredictive cue colour, F(1, 19)0 3.828, MSE027468.83, p0.065. On trials when cues were longer, RTs were shorter when the cue colour matched the target compared to when the cue colour differed; no difference was evident following shorter cues (100 ms cues, a difference of 53 ms, p0.376; 1200 ms cue, an effect of 198 ms, p0.009). Data 278 ANDERSON, HEINKE, HUMPHREYS from invalid trails demonstrated no reliable main effects or interactions (ps.07). RTs across Experiments 2 and 3. To examine how changing the predictive probability of the ‘‘other feature’’ influenced its effect on search, we compared valid and invalid RTs following 100 ms orientation-predictive cues in Experiment 3 with the comparative data from Experiment 2. For parsimony, only significant main effects and interactions involving experiment number will be reported. A mixed-design, three-factor ANOVA (experiment number, orientation validity, status of nonpredictive colour) revealed an interaction between the status of the nonpredictive cue colour and experiment, F(1, 36)06.686, MSE091992.54, p0.014, as well as a significant three-way interaction, F(1, 36)06.097, MSE063390.567, p0.018. Unpacking the latter interaction, RTs were shorter in Experiment 2 compared to Experiment 3 when both cue orientation and cue colour matched the target (a difference of 491 ms, pB.001). No other comparisons reached significance (ps.1). Accuracy. There was no evidence of a speedÁaccuracy tradeoff. Mean accuracy is shown in Table 2. Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 Discussion As in Experiment 2, there was a robust effect of the colour of the cue on search RTs when the orientation of the cue correctly predicted the target, with a benefit when the cue and target shared the same colour compared to when they differed. The cue’s colour affected search behaviour here even though it was contrapredictive, only matching the target on 20% of trials, with search facilitated when cue orientation matched the target compared to when the two differed. If anything this effect increased with longer cue duration, suggesting participants were unable to inhibit the effects of cue colour even with more time available. Although these findings suggest that the colour of a visual cue affects RTs automatically, reflexively directing attention towards search stimulus matching it in colour (cf. Posner et al., 1982), the effect decreased when the predictive feature (orientation) was invalid (comparing Experiments 2 and 3), providing evidence that an additional explicit component can also moderate search. Alternatively, a somewhat different account of performance may be considered. Given that the colour and orientation values of targets covaried, it is possible that participants generate an expectancy of the other property of the target from the predictive feature of the cue (e.g., given a blue cue then the target’s horizontal orientation could be predicted). The orientation of the target would be implicitly cued via its association with the cue colour (see Huettig & Altman, 2005; Moores et al., 2003). Therefore, the effects of the nonpredictive attributes of cues in IMPLICIT AND EXPLICIT CUEING IN VISUAL SEARCH 279 Experiment 2 may reflect a combination of the effects of top-down and bottom-up implicit processes. When the two factors are in agreement, overall implicit effects would be maximized. Whichever account of search is considered, the results of Experiment 3 suggest a level of automatic processing of cue colour. However, what level of visual processing is the implicit information affecting? Participants could be coding the cue as an object, with the unattended feature transferred into working memory in conjunction with the attended cue information (cf. Soto & Humphreys, 2009). Previous studies have proposed that cueing information about the target is transferred into working memory, guiding search in this manner (e.g., Huettig & Altmann, 2005; Moores et al., 2003). Therefore, participants could have retained the predictive task-relevant cue information in working memory, with the nonpredictive feature coded as part of the process. The features of the whole cue would affect RTs, whether they are attended to or not. Alternately, the nonpredictive colour of the cues in Experiment 3 could have affected search by priming the visual system to expect a target with the same colour, in a similar fashion to the facilitation evident from repeating a target feature trial-on-trial (e.g., Becker & Horstmann, 2009; Kristjansson, 2006; McBride, Leonard, & Gilchrist, 2007). Indeed, a carryover ´ effect of target colour has been demonstrated when colour was irrelevant to the search (Kristjansson, 2006). ´ The effects of cue colour in Experiment 3, however, were dependent on the validity of the predictive feature. When the orientation of the cue was valid, there was an added facilitation when the colour also matched the target relative to when it did not. In contrast, when cue orientation was invalid effects of cue colour were much reduced. If the implicit effects of the cue were automatic (and similar to the processes following explicit cue information), then, following an invalid cue (e.g., the orientation of the cue would be invalid), the colour of the cue would initially guide search towards stimuli sharing the colour of the cue (cf. Becker & Horstmann, 2009). If the colour of the cue and target matched, this process would oppose and lessen the effect of the invalid feature of the cue. A mismatch would increase the effect of the invalid predictive information in slowing search. Similar effects would be expected during the later stages of search, where attention may be reoriented towards stimuli not matching the predictive feature of the cue (e.g., orientation), but included in the target-containing subset. A match between the cue’s colour and the target would guide attention towards the target compared to when the cue and target colour differed. If the effects from the cued colour on invalid trials were comparable to those on valid trials, similar priming effects would also be expected, e.g., facilitation when the cue and target match in colour compared to when they differed. This was not reflected in the data. Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 280 ANDERSON, HEINKE, HUMPHREYS Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 One proposal is that the effect of cued colour degrades during the longer RTs of invalid trials (cf. Vickery et al., 2005; Wolfe et al., 2004). Priming would have only affected the early stages of search, with less effect later in the process (e.g., when attention is reoriented to the target-containing subset of uncued stimuli). Implicit effects of the colour of invalid cues would therefore have a reduced effect on RTs compared to when the cues were valid. A similar temporal argument could be made for the implicit effects arising from retaining the cue stimulus in working memory (cf. Soto, Rotshtein, Hodsoll, & Humphreys, 2008). Implicit effects from the colour of a cueing stimulus retained in working memory may be short-lasting (Soto & Humphreys, 2009), with colour having no influence at longer SOAs (see Olivers et al., 2006). GENERAL DISCUSSION The three experiments reported here show that the explicit effects of precueing the target in a visual search task were influenced by the physical nature of the cue. In all experiments, a visual cue predicted the colour or orientation of the following target in conjunction search, with the cueing information having 80% validity. Experiment 1 established a baseline where both physical and predictive properties from cues were the same (e.g., no implicit information), varying the cue duration from short (100 ms) to long (1200 ms) cues. Search was facilitated when cues matched the target compared to trials on which the cue and target did not match. There was a larger modulation of attention when colour rather than orientation cues were given. RTs were shorter following valid colour compared with valid orientation cues at all cue durations; however, search was slowed to a greater extent following invalid colour cues only when cues were presented for 200 ms (e.g., not following 100 ms or 1200 ms cues). The data show stronger guidance when the colour of the cue matched the target*valid cues increased the frequency of early fixations being made to the target compared to invalid cues. However, it is difficult to delineate whether the advantage was due to priming from the physical properties of the cue or differences in the processing of the predictive cue information. In Experiment 2, predictive information related to the target was again presented visually prior to search; however, a secondary cue feature was introduced that randomly matched the target. In one experimental session, the colour of the cue (a coloured bar) was predictive but its orientation was not, whereas the opposite was true in an alternate session (e.g., the orientation of the line was predictive, the colour nonpredictive). There were sizeable implicit effects from this ‘‘other’’ nonpredictive feature, especially when there was a valid orientation cue and the nonpredictive feature was IMPLICIT AND EXPLICIT CUEING IN VISUAL SEARCH 281 Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 colour. At both cue durations (100 ms and 200 ms), a strong facilitation was evident when there was a match between the nonpredictive feature and the target, compared to when there was a mismatch. The implicit effects from cue colour brought performance with orientation-predictive cues to that found with colour-predictive cues alone. This suggests that the stronger effect of visual colour (see Experiment 1) cues could be a consequence of automatic priming rather than top-down processing of the cue (cf. Anderson et al., 2010). In Experiment 3, similar effects of the nonpredictive colour of the cue were still evident when the nonpredictive attribute matched the target on only 20% of trials, whereas the predictive feature (orientation) remained at 80% validity. Cued colour affected search following both short and long cues; however, the effects were only reliable when the orientation of the cue was valid, suggesting that the implicit effects may not be automatic. There was a reduction in the implicit effects of colour when it was contrapredictive compared to when cue colour matched the target at chance, revealing that a level of strategic processing was present when the colour of the cue randomly matched the target (on 50% of trials, in Experiment 2). Guidance from the cue One way to conceive how guidance operates following featural cueing of a conjunction target is that cues direct search towards the cued stimuli matching the cued feature (see Anderson et al., 2010). If the target falls within this subset (e.g., on valid trials), search will be facilitated compared to if the target does not (e.g., on invalid trials). It may be assumed that predictive and nonpredictive information from the cue guides search via a similar mechanism. An alternate proposal is that within the current dualtarget methodology, search is executed first for one target and then, if needed, the other (Wolfe, 1992). The cue may therefore bias search towards a particular target template and, following the selection of the template, a typical guided search for the corresponding target would occur. Valid cues would facilitate performance as search for only one template would be necessary, whereas invalid cues may lead to a more exhaustive search pattern. If cueing were to operate in this manner, one would expect minimal differences between the cueing of colour and orientation of the target. This was not the case in the current set of experiments nor in Anderson et al. (2010), which largely replicated methodology of Experiment 1 here. Stronger effects from colour cues*whether explicitly or implicitly linked to the target*were evident. 282 ANDERSON, HEINKE, HUMPHREYS Colour advantage The results revealed similar stronger implicit guidance from the colour of a cue when there was a predictive link between the colour of the target and cue. The implicit effects remained when the colour of the cue was contrapredictive (Experiment 3), but there was then a reduced effect compared to when the colour and target matched at chance (Experiment 2). This suggests the advantage for colour cueing found previously (Anderson et al., 2010) can be explained, at least partly, by differences in how the physical nature of the cue is processed. Colour has been shown to be perceived earlier than orientation (Moutoussis & Zeki, 1997), offering evidence towards an early benefit for guidance following a colour cue. Indeed, stronger guidance from cue colour, particularly following valid cues, was manifest following the shortest cue duration (100 ms). The level of processing following the early detection of colour information, however, and how this is translated into guidance, is not clear. Colour priming has been shown to direct eye movements (Becker & Horstmann, 2009), with effects from colour even when irrelevant to the search task (Kristjansson, 2006; McBride et al., 2007). The robust advantage for ´ implicit and explicit cueing for colour is consistent with increased bottom-up priming of the visual system enabling more efficient and speedy processing of the relevant stimuli (see Becker & Horstmann, 2009). An alternate proposal, though, is that guidance following cues is a consequence of the information being coded into working memory prior to the search (cf. Moores et al., 2003). Featural information maintained in working memory has been shown to affect RTs with a stronger effect from colour than shape cues (Soto et al., 2005). There is also evidence of an effect from the colour of a cue when colour was not the feature that had to be memorized but was present for the to-bememorized shape (Soto & Humphreys, 2009). It is possible that the larger explicit and implicit effects from the colour of the cue could involve both priming processes and working memory. It should be noted that we cannot definitively state that effects on search from predictive and nonpredictive cue colour involve the same mechanisms. Eye movement data (see Experiment 1c) suggest that predictive information from cue colour guides search but we would require similar data from nonpredictive features to affirm the proposal (cf. Becker & Horstmann, 2009). Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 Invalid cues Whatever the mechanism for the colour advantage, the implicit priming effects were reduced following invalid cues compared to the effects on valid trials (Experiments 2 and 3). This suggests that the effects either decrease IMPLICIT AND EXPLICIT CUEING IN VISUAL SEARCH 283 during the longer RTs on invalid trials, are contingent on the validity of the predictive cue feature, or a combination of these factors. The temporal proposal fits with implicit processing occurring via mechanisms involving either priming or working memory, both of which have shown to be short lived (priming, Vickery et al., 2005; Wolfe et al., 2004; working memory, Olivers et al., 2006; Soto & Humphreys, 2009). On the other hand, it may be that during invalid trials, participants, after searching the cued subset, become aware of the invalid nature of the cueing information (cf. Posner et al., 1982) and use this knowledge to reorient attention towards the targetcontaining subset. Participants therefore reject the predictive information from the cue in response to the absence of the cued target. If the features of the cue were coded together (cf. Soto & Humphreys, 2009), the effect of the nonpredictive information would also be discarded. These two proposals could be tested by varying the number of stimuli in the search task, thereby either increasing the length of valid trials or reducing the length of invalid trials (cf. Anderson et al., 2010). The temporal hypothesis would suggest that implicit effects from the colour of the cue would negatively correlate with the number of search items (e.g., decreasing with increasing display size). Similar effects across array sizes, however, would offer evidence that participants reject all information once the cues have been identified as invalid, regardless of the time taken to complete the search. This remains to be tested. A further hypothesis would be that, whereas stimulus-driven implicit effects from the colour of the cue are contingent on the cue’s validity, reduced effects on invalid trials may be due to competition with implicit effects from the feature associated with the predictive cue information. For example, an invalid horizontal cue would also implicitly prepare the visual system for the expected blue target, even though the present target would be green. Therefore, if the cue was green, any priming effects from this matching the target (e.g., Becker & Horstmann, 2009) may be reduced as these processes would clash with the blue information associated with the predictive horizontal feature. On this view, stimulus-driven priming may be automatic but diminished on invalid trials due to competition with topdown implicit processes triggered via the target template. Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 CONCLUSIONS The experiments confirmed that there are stronger effects of colour than orientation cues on target detection in search tasks, with the cues influencing how efficiently attention is guided to targets. At least part of this effect is due to automatic, implicit cueing of attention from the physical properties of the cue, in addition to any greater top-down effects from an expectancy for the target feature. 284 ANDERSON, HEINKE, HUMPHREYS REFERENCES Anderson, G. M., Heinke, D., & Humphreys, G. W. (2010). Featural guidance in conjunction search: The contrast between orientation and color. Journal of Experimental Psychology: Human Perception and Performance, 36(5), 1108Á1127. Bacon, W. F., & Egeth, H. E. (1997). Goal-directed guidance of attention: Evidence from conjunctive visual search. Journal of Experimental Psychology: Human Perception and Performance, 23, 948Á961. Becker, S. I., & Horstmann, G. (2009). A feature-weighting account of priming in conjunction search. Attention, Perception, and Psychophysics, 71(2), 258Á272. Duncan, J., & Humphreys, G. W. (1989). Attentional engagement theory: Visual search and stimulus similarity. Psychological Review, 96(3), 433Á458. Goolsby, B. A., & Suzuki, S. (2001). Understanding priming of colour-singleton search: Roles of attention at encoding and ‘‘retrieval’’. Perception and Psychophysics, 63, 929Á944. Hannus, A., van den Berg, R., Bekkering, H., Roerdink, J. B. T. M., & Cornelissen, F. W. (2006). Visual search near threshold: Some features are more equal than others. Journal of Vision, 6, 525Á540. Huang, L., Holcombe, A. O., & Pashler, H. (2004). Repetition priming in visual search: Episodic retrieval, not feature priming. Memory and Cognition, 32(1), 12Á20. Huettig, F., & Altmann, G. T. M. (2005). Word meaning and the control of eye fixation: Semantic competitor effects and the visual world paradigm. Cognition, 96, B23ÁB32. Ishihara, S. (1981). Ishihara’s tests for colour-blindness (24 plate ed). Tokyo, Japan: Kenehara Trading. Kean, M., & Lambert, A. (2003). The influence of a salience distinction between bilateral cues on the latency of target-detection saccades. British Journal of Psychology, 94, 373Á388. Kristjansson, A. (2006). Simultaneous priming along multiple feature dimensions in a visual ´ search task. Vision Research, 46, 2554Á2570. ´ Kristjansson, A., Wang, D., & Nakayama, K. (2002). The role of priming in conjunctive visual ´ search. Cognition, 85, 37Á52. Kumada, T. (2001). Feature-based control of attention: Evidence for two forms of dimension weighting. Perception and Psychophysics, 63, 698Á708. Maljkovic, V., & Nakayama, K. (1994). Priming of pop-out. 1. Role of features. Memory and Cognition, 22(6), 657Á672. McBride, J., Leonard, U., & Gilchrist, I. D. (2007). Flexible target representations underlie priming in visual search. Perception, 36(2), 308Á308. Meeter, M., & Olivers, C. (2006). Intertrial priming stemming from ambiguity: A new account of priming in visual search. Visual Cognition, 13, 202Á222. Moores, E., Laiti, L., & Chelazzi, L. (2003). Associative knowledge controls deployment of visual selective attention. Nature Neuroscience, 6(2), 182Á189. Moutoussis, K., & Zeki, S. (1997). Functional segregation and temporal hierarchy of the visual perceptive systems. Procedures of the Royal Society of London, 264B, 1407Á1414. Muller, H. J., Krummenacher, J., & Heller, D. (2004). Dimension-specific inter-trial facilitation ¨ in visual search for pop-out targets: Evidence for a top-down modulable visual short-term memory effect. Visual Cognition, 11, 577Á602. Muller, J. M., Reimann, B., & Krummenacher, J. (2003). Visual search for singleton feature ¨ targets across dimensions: Stimulus and expectancy-driven effects in dimensional weighting. Journal of Experimental Psychology: Human Perception and Performance, 29(5), 1021Á1035. Neely, J. D. (1977). Semantic priming and retrieval from lexical memory: Roles of inhibitionless spreading activation and limited-capacity attention. Journal of Experimental Psychology: General, 106(3), 226Á254. Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 IMPLICIT AND EXPLICIT CUEING IN VISUAL SEARCH 285 Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 Olivers, C. N. L., & Meeter, M. (2006). On the dissociation between compound and present/ absent tasks in visual search: Intertrial priming is ambiguity driven. Visual Cognition, 13(1), 1Á28. Olivers, C. N. L., Meijer, F., & Theeuwes, J. (2006). Feature-based memory-driven attentional capture: Visual working memory content affects visual attention. Journal of Experimental Psychology: Human Perception and Performance, 32, 1243Á1265. Posner, M. I., Cohen, Y., & Rafal, R. D. (1982). Neural systems control of spatial orienting. Philosophical Transactions of the Royal Society of London, 298B, 187Á198. Schneider, W., Eschman, A., & Zuccolotto, A. (2002). E-Prime reference guide. Pittsburgh, PA: Psychology Software Tools, Inc. Soto, D., Heinke, D., Humphreys, G. W., & Blanco, M. J. (2005). Early, involuntary top-down guidance of attention from working memory. Journal of Experimental Psychology: Human Perception and Performance, 31(2), 248Á261. Soto, D., & Humphreys, G. W. (2009). Automatic selection of irrelevant object features through working memory evidence for top-down attentional capture. Experimental Psychology, 56(3), 165Á172. Soto, D., Rotshtein, P., Hodsoll, J., & Humphreys, G. W. (2008). Automatic guidance of attention from working memory. Trends in Cognitive Sciences, 12, 342Á348. Theeuwes, J., Reimann, B., & Mortier, K. (2006). Visual search for featural singletons: No topdown modulation, only bottom-up priming. Visual Cognition, 14, 466Á489. Van Zoest, W., Donk, M., & Theeuwes, J. (2004). The role of stimulus-driven and goal-driven control in saccadic visual selection. Journal of Experimental Psychology: Human Perception and Performance, 30(4), 746Á759. Vickery, T. J., King, L.-W., & Jian, Y. (2005). Setting up the target template in visual search. Journal of Vision, 5, 81Á92. Williams, D. E., & Reingold, E. M. (2001). Preattentive guidance of eye movements during triple conjunction search tasks: The effects of feature discriminability and stimulus eccentricity. Psychonomic Bulletin and Review, 8, 476Á488. Wolfe, J. M. (1992). ‘‘Effortless’’ texture segmentation and ‘‘parallel’’ visual search are not the same thing. Vision Research, 32(4), 757Á763. Wolfe, J. M. (1998). Visual search. In H. Pashler (Ed.), Attention (pp. 13Á74). London, UK: University College London Press. Wolfe, J. M., Horowitz, T. S., Kenner, N., Hyle, M., & Vasan, N. (2004). How fast can you change your mind? The speed of top-down guidance in visual search. Vision Research, 44, 1411Á1426. Manuscript received June 2009 Manuscript accepted September 2010 286 ANDERSON, HEINKE, HUMPHREYS APPENDIX: EQUATING SEARCH EFFICIENCIES FOR THE STIMULI USED IN EXPERIMENT 1C In Experiment 1c, changes were necessary in display density and presentation set-up from that used in the rest of this paper to allow for accurate eyetracking. This dictated that further piloting of stimuli was required to control for bottom-up dimensional differences in search efficiency (see Anderson et al., 2010, for details of pilot studies of all other experiments). To avoid differential top-down cueing effects, participants performed an ‘‘odd-one out’’ task in which they searched for a feature singleton that could occur along either the colour or orientation dimension. If the saliencies of the targets along each dimension are equated, there should be no search advantage for colour- over orientation-defined targets (Bacon & Egeth, 1997). Participants also performed a conjunction search task with the same stimuli to check whether search was then inefficient. Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 Method Unless otherwise mentioned, this matched that of Experiment 1c. Participants. Ten University of Birmingham students, four male, six female, aged between 23 and 28 (average age 25.4) took part. Stimuli. We undertook exploratory experiments varying the colour saturation finally settling on the colour levels shown in Table 1, which were used for this study. Design. For the feature-singleton search there were two main independent variables: The defining dimension (colour, orientation) and target presence (present, absent); there was only one main independent variable for the conjunction search task (array size) as the response depended on the symbol on the search target rather than whether it was present or absent. Procedure. Half the participants performed the feature-singleton task, followed by the conjunction task; the order was reversed for the other participants. In feature-singleton tasks, participants were instructed to search for a stimulus that was the odd-one out from the array. First, a fixation circle was present for 1000 ms, then a 100 ms interstimulus interval (ISI) before an array of stimuli with one target and eight distractors. The defining dimension and target for each array was varied trial-on-trial in equal numbers. For colour-defined search, targets and distractors possessed different colours yet the same orientation (blue horizontal target vs. green horizontal distractors, IMPLICIT AND EXPLICIT CUEING IN VISUAL SEARCH 287 or green vertical target vs. blue vertical distractors). For orientation-defined search, the targetÁdistractor relationship differed only along the orientation dimension (blue horizontal target vs. blue vertical distractors; or green vertical target vs. green horizontal distractors). When the target was absent (50% of the time), it was replaced by a distractor. Participants undertook 24 practice trials followed by two blocks of 80 experimental trials. The methodology for the conjunction task was identical to that of Experiment 1c, except that no cues were presented prior to each trial. There were 18 practice trials followed by two blocks of 48. Only RT and accuracy were recorded. Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 Results RTs. Data were cleaned as previously. For the feature-singleton task, the data were pooled across trials within the target-defining dimension (colour or orientation) and median RTs for each participant were calculated. Targetabsent trials were treated as catch trials, so only target-present data were 700 Colour-defined search 2100 Orientation-defined search Conjunction search 680 2000 1900 1800 1700 1600 1500 1400 1300 1200 1100 RTs (ms) from feature-singleton task 660 640 620 600 580 560 540 520 500 5 9 Array size Figure 7. Means ('/( one standard error) of median RTs from the pilot study outlined in the Appendix, separated by search type (colour-, orientation-, or conjunction-defined task) and array size (note only the conjunction-defined task included a five-item array). RTs (ms) from conjunction task 288 ANDERSON, HEINKE, HUMPHREYS analysed. The means of target-present median RTs for both feature-singleton and conjunction search are shown in Figure 7. A one-factor ANOVA showed no difference between RTs from colourand orientation-defined displays (FB1). Data from the conjunction task were pooled across target type and were also analysed using a one-factor ANOVA (array size). The data were typical of an inefficient search task (Wolfe, 1998), with a borderline significant main effect of array size, F(1, 9)04.607, p0.06, partial h2 0.339, indicating an increase in RTs as array size increased. The search slope was 45 ms/item. Accuracy. In both search tasks, performance followed that of the RT data indicating there was no speedÁaccuracy tradeoff. Errors are presented in Table 2. Downloaded By: [University of Birmingham, UK] At: 13:29 4 March 2011 Discussion The data confirm that (1) the orientation-defined targets used here were as salient as the colour-defined targets, and (2) search for conjunction targets was relatively inefficient. Given these patterns of search, the same feature values were used in Experiment 1c.