Vision Research 49 (2009) 1174–1181

Contents lists available at ScienceDirect

Vision Research

journal homepage: www.elsevier .com/locate /v isres

Maximal motion aftereffects in spite of diverted awareness

Erik Blaser *, Tim ShepardDepartment of Psychology, University of Massachusetts Boston, 100 Morrissey Blvd. Boston, MA 02125, USA

a r t i c l e i n f o

Article history:Received 4 October 2007Received in revised form 19 August 2008

Keywords:AttentionAwarenessMotion aftereffectAuditoryDark attention

0042-6989/$ - see front matter � 2008 Elsevier Ltd. Adoi:10.1016/j.visres.2008.09.012

* Corresponding author.E-mail address: erik.blaser@umb.edu (E. Blaser).

a b s t r a c t

This study attempts to isolate the underlying processing resources of visual attention from the ‘cognitivesupervision’—working memory, decision processes, but especially awareness—that typically accompaniestheir allocation. To decouple them, we used the motion aftereffect (MAE) as a passive assay of resourceallocation. In our main condition, observers were presented with an adapting field, but did not attend toit. Instead their effort was directed to an engrossing auditory two-back memory task. Consequently,observers had no consistent awareness of the adaptor, nor were able to make accurate judgements aboutits luminance, but nonetheless had MAE’s no smaller than those induced when the adaptor was ‘fullyattended’. Similarly to when object- or feature-based attention spreads unwittingly, attention was allo-cated automatically to the adaptor, without requiring nor engaging executive control or awareness.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction

At least since William James described attention as ‘‘...the takingpossession by the mind, in clear and vivid form, of one out of whatseem several simultaneously possible objects or trains of thought.Focalization, concentration, of consciousness are of its essence....”(James, 1890),‘attention’—the selective enhancement and prioriti-zation of behaviorally relevant information (Treue, 2003)—and‘awareness’—the real-time, conscious experience of a stimulus—have been tightly entwined. Though the two have been distin-guished (Bahrami, Carmel, Walsh, Rees, & Lavie, 2008; Hardcastle,1997; Kentridge, Heywood, & Weiskrantz, 2004; Koch & Tsuchiya,2007; Naccache, Blandin, & Dehaene, 2002), it can still seem a par-adox to speak of attending to something without being aware of itor, vice versa, to have a proximal stimulus enter awareness with-out attending to it (Dehaene, Changeux, Naccache, Sackur, & Ser-gent, 2006; He, Cavanagh, & Intriligator, 1996; Lamme, 2003;Merikle & Joordens, 1997; O’Regan, & Noe, 2001; Posner,1994; Vel-mans, 1996). This though leads to a contradiction. If attention andawareness are coupled, this implies visual awareness too must benecessary for the processes for which attention is thought neces-sary, such as ‘object binding’ (Treisman & Gelade, 1980), memorygating (Sperling, 1960), ‘serial’ visual search (Treisman & Gelade,1980; Wolfe, Cave, & Franzel, 1989), and much of object detection(where attention misallocation is thought to induce ‘inattentionalblindness’ (Mack & Rock, 1998) and ‘change blindness’ (Rensink,O’Reagan, & Clark, 1997; Simons & Levin, 1997), just to name afew. In short this means without visual awareness we would be

ll rights reserved.

effectively blind; yet many regular commuters, lost in thought, ar-rive home without incident.

A relatively new line of experimental evidence also points to thenecessity of better distinguishing attention from awareness. Inthese attention-spreading effects, performance or responses ofneurons is not just affected at explicitly attended locations (Con-nor, Preddie, Gallant, & Van Essen, 1997), but also winds up af-fected elsewhere unwittingly and automatically. For instance,consider ‘cross-attribute’ attention where, for example, attendingto the color of a field of drifting dots induces attention to their mo-tion (Sohn, Papathomas, Blaser, & Vidnyanszky, 2004), ‘feature-based attention spreading’, where attention to a feature at onelocation induces global attention effects to like-features beyondthis intended focus, for instance when attention to motion in onelocation influences the responses of similarly tuned neurons inother locations (Arman, Ciaramitaro, & Boynton, 2006; Melcher,Papathomas, & Vidnyanszky, 2005; Saenz, Buracas, & Boynton,2002; Serences & Boynton, 2007; Treue & Martinez Trujillo,1999; Treue & Maunsell, 1996), and ‘object-based attention’ whereattention devoted to one feature of an object triggers attention toall of its concomitant features (Blaser, Pylyshyn, & Holcombe,2000; Duncan, 1984; O’Craven, Downing, & Kanwisher, 1999); allof this shows that there is much more to attention than the thin ve-neer of what we direct it to do, and what we are aware of it havingdone.

Our goal here is to divorce visual processing resources from theexecutive, central bottleneck processes (Pashler & Jhonston,1998)—working memory, decision, but especially awareness—thattypically accompany their allocation. In this way we can observehow resources are allocated when unsupervised. In practice it isnotoriously difficult to tease the two apart (‘‘Please attend solely

E. Blaser, T. Shepard / Vision Research 49 (2009) 1174–1181 1175

to the red dots on the left while keeping your awareness just on thegreen dots on the right”). To measure resource allocation withoutinfluencing it, we used the magnitude of the motion aftereffect(MAE) as a passive assay.

1.1. Attention and the MAE

In 1911, Wolhgemuth had observers view a moving stimulus inhis effort to induce and quantify the duration of the MAE (Wohlge-muth, 1911), an illusion of movement now thought to reflect animbalance in the relative responses of direction-selective corticalneurons induced by prolonged viewing of an adapting motionstimulus (Mather, 1980; Sutherland, 1961; Mather and Harris,1998). He noted significant MAE’s. In a separate condition designedto assess the influence attention on this effect, he had them againview the adapting motion stimulus, but asked them to concentrateon a mental arithmetic task. He noted no change in MAE durations.However, nearly a century later, Chaudhuri (1990) performed asimilar study, with very different results. In a condition whereobservers diverted their attention to an RSVP (Sperling, Budiansky,Spivak, & Johnson, 1971) task at fixation he found drastic—on aver-age 50%—reductions in MAE durations relative to when observerspassively viewed the adapting field of dots (Fig. 1).

A large body of subsequent behavioral (Alais & Blake, 1999;Georgiades & Harris, 2000; Lankheet & Verstraten, 1995; Rezec,Krekelberg, & Dobkins, 2004; von Grünau, Bertone, & Pakneshan,1998) and neurophysiological (Berman & Colby, 2002; Huk, Ress,& Heeger, 2001; Rees, Frith, & Lavie, 1997; Seiffert, Somers, Dale,& Tootell, 2003) studies have confirmed Chaudhuri’s (1990) basicresult: attending to an adapting motion stimulus maximizesMAE’s, while diverting it minimizes them.

1.2. Isolating ‘dark attention’

The term ‘attention’ is imprecise. Some visual processing re-sources may be willfully directed, as when the observer executesan instruction or implements a top-down allocation strategy, butthere are circumstances—the vast majority of circumstances, wewould argue—when there is a context-dependent, flexible shut-

Fig. 1. Main results reprinted from Chaudhuri (1990). This graph shows MAEduration for five observers under two attentional conditions: a passive conditionwhere observers simply viewed the motion adaptation stimulus (a laterally driftingnoise field) and a withdrawn attention condition where observers were asked toperform an RSVP task at fixation during adaptation. In this condition, MAEdurations dropped on average 50%. We replicate this main effect, but using a truedual-task paradigm, in Experiment 1.

tling of resources between visual stimuli, but that neither re-quires nor engages executive control or awareness. We refer tothis as the shadow economy of ‘dark attention’. This is not a pro-posal of new mechanism of attention per se, but an explicit dis-tinction between cases where resources are willfully allocated,and cases where allocation occurs unsupervised. In our mainexperiment, we try to divert only the panoply of cognitive andconscious processes away from a visual stimulus, to removesupervision from the underlying processing resources, therebyisolating dark attention.

In our main conditions observers were presented with anadapting field of moving dots while performing two tasks con-currently (Braun & Sagi, 1990; Sperling & Melchner, 1978). Onetask required observers to make judgments about the luminanceof the adapting field, while the other required observers to mon-itor a stream of numerals to perform a two-back memory task.Observers were instructed how to distribute their effort betweenthe two tasks (e.g. ‘‘give 90% of your effort to the two-back task,10% to luminance”). These instructions were designed to inducetradeoffs in awareness through concomitant, measurable trade-offs in task-relevant cognitive processes, such as workingmemory.

Importantly, we presented the two-back task in two differentmodalities. In our main experiment, the two-back task was pre-sented auditorially (a spoken stream of numerals presented overheadphones) to eliminate demands on visual processing resources(Duncan, Martens, & Ward, 1997; Treisman & Davies, 1973) whilemaximizing cognitive demands (Brand-D’Abrescia and Lavie,2008). In a contrasting experiment designed as a basic replicationof Chaudhuri, the two-back task was presented visually (an RSVPstream of numerals presented at fixation), thereby invoking signif-icant, likely zero-sum visual demands. In both experiments, weused the magnitude of the motion aftereffect (MAE) as a passiveassay resources allocated to the adapting field.

We found that instructions to weight one task or the other didindeed induce tradeoffs in performance between the two tasks.Critically, this was true for both the visual version and the audi-tory modalities. Given the previous results on attention and theMAE, it is tempting then to expect then that MAE’s would beweaker when observers’ effort was diverted from the luminancetask to (either of) the two-back tasks. Consistent with this predic-tion, MAE’s did drop considerably when effort was devoted pri-marily to the visual two-back task at fixation, away from theluminance task on the adapting field. However, in sharp contrast,MAE’s were undiminished when effort was directed to the audi-tory two-back task: In a condition where observers ignored theadapting field and instead devoted themselves to an engrossingalternate task—and as a consequence had no consistent aware-ness of the adapting field and no ability to make judgments aboutits luminance—MAE’s nonetheless were no smaller than those in-duced in a condition where the adaptor was fully ‘attended’. Wetake this as evidence that a resource—dark attention—was none-theless allocated to the adaptor, automatically and unsupervised,supporting normal motion processing and therefore maximalMAE’s.

2. Methods

2.1. Observers

Six observers (4 naive and 2 expert) with normal or corrected-to-normal vision participated in these experiments. Informed writ-ten consent was obtained from all observers. Naive observers wereUniversity of Massachusetts Boston undergraduate students, whowere paid $8.00/h for their participation.

1176 E. Blaser, T. Shepard / Vision Research 49 (2009) 1174–1181

2.2. Apparatus

Stimuli were displayed on a ViewSonic PF790 monitor run-ning at 640 � 480 and 120 Hz. Presentation and response col-lection were controlled with VisionShell software running onan Apple Mac G4. Observers sat in a darkened room andviewed the monitor from a distance of 57 cm, with head posi-tion stabilized by a chin rest. Depending on condition, somestimuli were presented visually and/or auditorially, throughheadphones. Observers made responses through keystrokesand/or spoken responses into a microphone positioned underthe chin rest.

2.3. MAE stimuli

Stimuli consisted of 512 small (2 arc min) dots randomly dis-tributed on a dark background. Dot patterns were presented inan aperture with a diameter of 11 deg, yielding a dot densityof 5 dots per sq. deg. During stimulus presentation, observerswere instructed to fixate on a small central cross. During adap-tation, stimuli drifted laterally with a speed of 3 deg/s and100% coherence, and a limited lifetime of 317 ms. The field of

Fig. 3. Trace of roving luminance used for plateau detection task. The luminance of thperforming the ‘plateau detection’ task, observers monitored this roving for flat spots - bplateaus in a typical 30 s trial.

Fig. 2. Example trial. Every trial in both Experiment 1 and 2 consisted of two main phaswere presented with a drifting field of limited lifetime dots. This field also had dynamicaDuring this phase, observers were asked to perform two tasks concurrently: a ‘plateau’ dechanges, see Fig. 3) and a two-back memory task (where observers monitored a streamauditorially, over headphones. At the end of the 30 s adaptation period (the first trial ofentered the test phase. In the test phase, observers were presented with a physically statfield reached perceptual standstill, and then pressed a button to initiate the next trial. Tduration and served as our measure of MAE strength, consistent with the measure used

dots also changed luminance dynamically—in unison, randomlybut smoothly—between a dim but supra-threshold 3 cd/m2 tothe monitor’s maximum luminance of 164 cd/m2 (please seeFig. 3 for an example trace; this dynamic luminance was rele-vant to one of our tasks, described below). Adaptation lasted60 s for the first trial of a block and 30 s for subsequent trials.When the adaptation period ended, there was a 500 ms blankinterval and then the test period began. The test stimulus wassimilar to the adaptation field, but the dots were now physicallystatic, infinite lifetime and displayed at a fixed mid-range lumi-nance of 84 cd/m2. During the presentation of the test stimulus,the observer was likely to experience an MAE. When an observerjudged that the MAE had ended and the field had reached a per-ceptual standstill, she pressed a key to terminate the trial. MAEduration was defined as the elapsed time from the beginning ofthe test interval to this keypress (Fig. 2).

Duration was used as a measure of MAE magnitude to matchthe measure used by Wohlgemuth (1911) and Chaudhuri (1990).(It is worth noting here that we ran pilot experiments to determineif the use of limited-lifetime or infinite-lifetime ‘static’ dots in ourtest stimuli affected our pattern of results. As we found no differ-ence, we used static fields; we have noticed that naive observers

e adapting dots varied randomly, but smoothly between 3 and 164 cd/m2. Whenrief periods when the dots stayed at a fixed luminance. There were roughly 7 such

es, adaptation and test. During the adaptation phase, observers fixated centrally andlly roving luminance, with all dots simultaneously randomly, but smoothly, varying.tection task (where observers monitored for brief pauses in the dynamic luminance

of numerals for two-back repeats) that was presented either visually, at fixation, orevery block was 60 s), a short blank period appeared for 500 ms, and then the trialionary field of infinite lifetime dots. Observers judged when the illusory drift of thishe time from the end of the blank field to this button press was taken as the MAEby Chaudhuri (1990) and Wohlgemuth (1911).

E. Blaser, T. Shepard / Vision Research 49 (2009) 1174–1181 1177

are more comfortable judging when static fields reach perceptualstandstill.)

2.4. Primary task: plateau detection (luminance judgments withrespect to the adapting field)

Plateau detection required observers to monitor the luminancechanges in the adapting field. Occasionally, the luminance of thefield would stop changing for a brief period, then resume roving;observers had to make speeded detections (through a keypress)of these subtle, periodic plateaus. Plateaus lasted 1.5–2.5 s, and oc-curred with ISI’s of 0.5–3.5 s, yielding about 7–8 plateaus in a 30 strial (Fig. 3).

A ‘hit’ was registered when the observer correctly identified aplateau during its presentation, whereas a response outside of thisinterval was coded as a ‘false alarm’. The failure to respond withinthis interval was coded as a ‘miss’ while ‘correct rejections’ wererecorded when observers made no response during the intervalsbetween plateaus. These rates were used to compute percent cor-rect performance. Observers received feedback only when a re-sponse was made (the fixation cross turned green for hits andred for false alarms). The frequency and duration of these plateaushad been previously calibrated in pilot experiments to yield a dif-ficult, engrossing task with performance levels on average at 50%correct (corrected for guessing) for naive observers with a fewblocks of practice. This task is performed on the adapting field itselfand is a dummy task of sorts, designed coax observers into fully‘attending’ to the adapting field.

2.5. Alternate task: visual or auditory two-back memory

The second task in this dual-task paradigm was two-back mem-ory. This task was performed over a stream of numbers that waspresented either visually (in RSVP fashion, at fixation) or auditori-ally (via spoken numerals over headphones). Apart from presenta-tion modality, task and parameters were identical. The numerals1–5 were used, and were presented in random order at a rate ofabout one numeral every 2 s (1850–2150 ms). Every time therewas a two-back repeat in the sequence—for instance of the numer-al 3 in the sequence ‘‘...2 1 3 3 4 3 5 4...”—observers had to make aspeeded response (by speaking the word ‘yes’; observers made ver-bal responses for both the visual and auditory versions of this task).Responses immediately following a repeated numeral, before theonset of the subsequent numeral, were coded as hits; other re-sponses were coded as false alarms. Failure to respond within thisinterval resulted in a miss, while correct rejections were codedwhen observers withheld responses in the interval following anon-repeated numeral. These rates were used to calculate percentcorrect measures of performance.

2.6. General procedure

All trials consisted of an adaptation and test phase. During adap-tation, observers were presented with the adaptation stimulus andperformed the plateau and/or the two-back task. In dual-task con-ditions, observers could be given one of the following instructionsto ‘weight’ one task more strongly than the other: Give 100% ofyour effort to the plateau task (single-task control); give 90% effortto plateau detection, 10% to two-back memory; give 50%/50% effortto each task; give 10% effort to plateau, 90% to two-back; give 100%effort to two-back (single-task control). (In the dual-task condi-tions, it is possible that there could be response interference be-tween the two tasks. To rule this out, we performed a simplecontrol where the plateau task was replaced with a structurallyidentical, but perceptually trivially easy, task: every time the rov-ing dots entered a plateau, the dots flashed red briefly. Observers

were no worse at either task when performed together as wheneach was performed alone—ruling out response-related interfer-ence). Instructions were blocked and observers performed twoblocks of 10 trials each. MAE durations were measured in the testphase of each trial, allowing us to measure how the instructedbiases between the two tasks would affect this measured MAE.This all amounted to three measurements made for a typicaldual-task trial: performance on the plateau task, performance onthe two-back task, and MAE duration. Experiments 1 and 2, de-scribed below, differed only in whether the two-back task was pre-sented visually or auditorially.

3. Results

3.1. Experiment 1: MAE’s induced during plateau detection or visualtwo-back tasks

This experiment was designed as a minimal replication of Chau-dhuri’s (1990) finding that withdrawing ‘attention’ from an adapt-ing stimulus significantly reduces the magnitude of the inducedMAE. For expediency, only the two single-task baseline conditionswere tested during the adaptation phase of each trial: the plateaudetection task alone during adaptation, or the two-back task aloneduring adaptation. Observers ran three 10 trial blocks of each ofthese conditions. In a confirmation of Chaudhuri’s main result, anaive and expert observer both showed significant reductions inMAE duration when effort was devoted to the visual two-back taskat fixation, relative to when effort was devoted to plateau detectionon the adapting field itself. Expert observer TS’s MAE dropped from15.74 to 5.47 s, while naive observer AB’s MAE dropped from 7.61to 5.74 s. Both the absolute durations of these MAE’s and the pro-portional reductions are comparable to those found by Chaudhuri(1990).

3.2. Experiment 2: MAE’s induced during dual-task: plateau detectionand auditory two-back

During the adaptation phase of each trial in this experiment,observers performed both the plateau task and the auditory two-back task concurrently. Instructions were given to observers tomanipulate the effort that they devoted to each of the two tasks.This resulted in a total of five conditions, which were presentedin different blocks and in random order (each block contained a to-tal of 10 trials, and observers ran two blocks of each condition).These five conditions were: 100% effort to plateau detection (sin-gle-task control), 90% effort to plateau detection and 10% effortto auditory two-back, 50/50, 10/90, and 100% auditory two-back(single-task control).

As hoped, there was a significant tradeoff in performance be-tween the two tasks as a function of pre-block instruction: observ-ers did better when giving a task more effort, and showed aninability to do both tasks at the same time without loss (with theexception of observer AB). While of course there are tasks whichdo not compete for resources (like walking and chewing gum)these two tasks do show significant tradeoffs in performance, asseen in the average and individual attention operating characteris-tics (AOC’s; Sperling & Melchner, 1978) (Fig. 4). (Unfortunately, ourdata is not fine-grained enough to allow us to make any strongstatements about whether these tradeoffs reflect a sharing of cog-nitive resources between the two tasks, or a switching of resources.What we can say is that performance on a trial-by-trial basis, al-most without exception, did not show the telltale negative correla-tions between performance on the two tasks that would beexpected with resource switching. While the lack of correlation isexpected of sharing models, this analysis cannot rule out switch-

TSAB

RJZK

plateau task performance (prop.correct)

performance iscorrectedfor guessingand averagedover6

observers

2-backonly

90/10

50/50

10/90

two-

back

task

per

form

ance

(pro

p.co

rrect

)

plateau only0

0.125

0.250

0.375

0.500

0.625

0.750

0 0.125 0.250 0.375 0.500 0.625 0.750

EBMK0

0.25

0.50

0.75

0

0.25

0.50

0.75

0

0.25

0.50

0.75

0

0.25

0.50

0.75

0

0.25

0.50

0.75

0

0.25

0.50

0.75

0 0.25 0.50 0.75 0 0.25 0.50 0.75

0 0.25 0.50 0.750 0.25 0.50 0.75

0 0.25 0.50 0.75 0 0.25 0.50 0.75

Fig. 4. Attention Operating Characteristics (AOC), Experiment 2. The left panel shows the average AOC’s for each of the six observers, with individual AOC’s shown in theright-hand panels. The x-axis shows proportion correct performance (corrected for guessing) in the plateau task and the y-axis shows performance for the auditory two-backtask. Data points on the axes themselves indicate performance for the single-task control conditions (‘plateau only’ and ‘two-back only’). Dotted lines show theoretical AOC’sfor tasks that compete completely with one another or that are independent. Observers in general showed considerable competition, as can be seen in the individual datapanels and in average performance.

1178 E. Blaser, T. Shepard / Vision Research 49 (2009) 1174–1181

ing. Resources could have been switched within a trial, and we donot have the event-by-event breakdown for performance on eachplateau and numeral to reveal this.)

In this fashion, this experiment fits the setup for classicChaudhuri (1990) results like those found in our Experiment 1. Insharp contrast however, we found that MAE’s were undiminishedeven as less and less effort was given to the plateau task on theadapting field, and were undiminished even in 100% two-back taskconditions, when effort and awareness were completely with-drawn from the adapting field (see Figs. 5 and 6).

When observers fully weighted the adapting dots (100% plateautask), average MAE across our six observers was 10.10 s, and theMAE in the fully withdrawn condition (100% two-back task) wasidentical, at 10.26 s.

3.3. Adaptation duration control

It is possible that the maximal MAE’s we found in Experiment 1reflect a ceiling effect of sorts. Perhaps ignoring the adapting dotsactually does affect motion processing and would reduce theMAE, but this withdrawal from the adaptor to the auditory taskwas somehow incomplete. If the MAE then were very quick to sat-urate then any ‘leakage’ back to the adaptor might be sufficient togenerate the maximal MAE’s we observed. We ran a control exper-iment just to be sure. In this control, we simulated leakage by mod-ulating the duration of adapting motion during the 30 s adaptationphase. In one condition, adapting motion periods were variable(ranging from 4000 to 500 ms) and were broken up by a fixed500 ms ISI periods of either static dots or a blank field (blank inter-vals provide an even stronger test here, as they allow for morepotential for MAE ‘storage’ over the ISI periods, see Verstraten,

Fredericksen, van Wezel, Lankheet, & van de Grind, 1996). In an-other condition, adapting motion durations were fixed at 500 msbursts, with variable ISI’s (ranging from 500 to 3000 ms) againeither of static dots, or a blank field.

In all cases, maximal MAE’s only occur with adapting motionclose to the full, unbroken 30 s, all but ruling out explanationsbased on leakage (Fig. 7).

4. General discussion

In this study we attempted to divorce the underling visual pro-cessing resources of attention from the cognitive and consciousprocesses—working memory, decision, but especially awareness—that typically initiate or result from their allocation. Even whensuch cognitive supervision is removed—say driving while day-dreaming—we argue that there remains a context-dependent, flex-ible shuttling of processing resources between visual stimuli; ashadow economy of what we term ‘dark attention’. For instancein our main experiment, observers ignored an adapting field ofmoving dots, and instead gave their effort to an engrossing audi-tory memory task. In spite of this, MAE’s were undiminished,showing that visual attention can indeed ‘go dark’: processing re-sources were automatically directed to the adapting field, withoutrequiring nor engaging awareness.

4.1. ‘Attention’ without awareness

Aspects of the interplay of attention and awareness are conten-tious (Koch & Tsuchiya, 2007; Lamme, 2003). It is beyond the scopeof the present experiment to settle this issue definitively, but wetake our results as strong evidence that processing resources can

Plateau only2-back only

Mean MAE duration

(seconds)

Observer

0

4

8

12

16

E B TS AB MK R J ZK

Fig. 5. MAE duration by observer, Experiment 2. This graph shows MAE durations for each of our six observers in single-task control conditions where observers only had toperform one task during a block of trials (error bars reflect standard errors). In the plateau only condition, observers devoted all their effort to the plateau task (which hadobservers continuously monitoring the roving luminance of the adapting field) and therefore to the adapting field itself. In two-back only conditions, observers were asked todevote all their effort to the auditory two-back memory task. This task competed significantly for cognitive and conscious resources with the plateau task (as results in Fig. 4show). Even under fully withdrawn conditions, when observers gave full effort to the two-back task and ignored the adaption motion field, MAE’s were undiminished.

0

4

8

12

16

platea

u90

/1050

/5010

/90

two-b

ack

EBABTSMKRJZK

MAE duration

(seconds)

Fig. 6. MAE duration as a function of attention allocation condition, Experiment 2.This graph shows that pattern of MAE duration, for each of our six observers, as afunction of the attention allocation instruction that they were given for a particularblock of trials (i.e. ‘give 100% of your attention to the plateau task’; ‘give 90% to theplateau task and 10% to the auditory two-back task’; ‘50/50’; ‘10/90’; and ‘100%auditory two-back’). As the graph shows, MAE’s were unaffected by theseinstructions, even though they resulted in dramatic shifts in performance andawareness between the two tasks (see Fig. 4).

E. Blaser, T. Shepard / Vision Research 49 (2009) 1174–1181 1179

be distributed independently of awareness. One can readily findother evidence, as in reports of cueing at work in individuals withblindsight (Kentridge et al., 2004), or the significant tilt aftereffectsinduced by an adapting patch that has been rendered indiscrimina-ble due to the crowding of flanking patches (He & Cavanagh, 1996;Montaser-Kouhsari & Rajimehr, 2005). Even in the relatively cir-cumscribed domain of attention and the MAE, there is convergingevidence to support this independence. For example, Blaser, Papa-thomas, and Vidnyanszky (2005) performed a study where theyshowed that color-contingent MAE’s were undiminished evenwhen the adapting motion was invisible (see also Whitney & Bress-ler, 2007). In one condition, the adapting stimulus contained arightward field of red dots and a leftward field of green dots, slid-ing transparently over one another. In another condition, dots were‘locally paired’ such that every rightward red dot was on a collisioncourse with a nearby leftward green dot; in this case the motion‘canceled out’ perceptually, and the adaptor appeared as a motion-less flicker. MAE’s were identical in both cases. This is strong evi-dence that the underlying processing resources—dark attention—

were allocated without direction, since in the invisible-adaptorcondition there was simply no relevant target for executive controlor awareness to latch onto (while in the transparent condition,either or both of the clearly visible adapting motion vectors couldbe engaged).

4.2. Competition for dark attention

To what extent dark attention influences, for instance, the MAEdepends in part on how much competition there is between theadapting field and other stimuli. Indeed, the magnitude of theMAE in our experiments was manipulated by the nature of analternate task. In some contexts there is competition, for instance,an RSVP task at fixation (our Experiment 1; Chaudhuri, 1990; Re-zec, 2004), where dark attention targets the RSVP characters atthe expense of the adapting field, resulting in reduced aftereffects.Similarly, in a recent study that follows from Lavie’s (1995) ‘per-ceptual load’ model of attention, Bahrami et al. (2008) found thatthe extent of adaptation to an oriented patch—which, critically,was masked from awareness—was determined by the perceptualdemands of a concurrent visual task. (It is important here to differ-entiate what we are calling dark attention from various invocationsof ‘preattentive’ or non-attentive visual processes (Braun & Julesz,1998; Braun & Sagi, 1990; Moray, 1959; Neisser, 1967; Schneider &Shiffrin, 1977; Treisman & Gelade, 1980; VanRullen et al., 2004;Wolfe et al., 1989). Such processes are understood as mandatory,while dark attention is flexible.)

Given this, alternate tasks in non-visual modalities (Duncanet al., 1997; Treisman & Davies, 1973) should not affect the MAE,as they will not compete for visual processing resources. This, ofcourse, explains the disparity in Wolhgemuth’s (1911) and Chau-dhuri’s (1990) results, as well as the difference in results betweenthe present study’s Experiment 1, which used a visually presentedtwo-back task, and Experiment 2, which used an auditory two-back task. There is additional converging behavioral and neuro-physiological evidence that shows greater modulation of theMAE and related cortical activity when ‘attention’ is diverted to avisual as opposed to an auditory task (Ciaramitaro & Boynton,2007; Rees, Frith, & Lavie, 2001). (However, this interaction withmodality is still not completely understood. Attention can modu-

0

6

12

18

base

line40

0035

0030

0025

0020

0015

0010

00 500

MAE

duration(seconds)

Adaptation duration (msec)

with 500 msec adaptation

ISI’s

0

6

12

18

base

line

500

1000

1500

2000

2500

3000

Blank fieldStatic field

Blank fieldStatic field

MAE duration

(seconds)

Adaptation ISI (msec)

with 500 msec adaptation durations

Fig. 7. Effect of adaptation duration on MAE duration. This control experiment was designed to test the possibility that our maximal MAE’s might reflect ‘leakage’. Under afast saturation model of the MAE, if the cognitive and conscious resources demanded by the two-back task do in fact influence the MAE, and if some of these resources ‘leak’back to the adaptor, large MAE’s could result. We tested this by simulating leakage, so to speak, by filling the adapting phase with periodic bursts of adapting motion, in otherwords simulating awareness as ‘turning on’ the adapting field. The duration of the adaptation phase was always 30 s, but was broken up into adaptation motion periodsversus ISI periods in two ways. The left panel shows MAE durations when the 30 s adaptation phase is broken up into variable-length periods of adapting motion (rangingfrom 4000 to 500 ms) separated by fixed 500 ms ISI’s of either a static field or a blank field; the results from blank field conditions are even a stronger test, as these potentiallyallow for more aftereffect ‘storage’ (Verstraten, et al., 1996) during the ISI. The right panel shows MAE durations when the 30 s adaptation interval is broken into fixed 500 msintervals of adapting motion separated by variable-length ISIs (of either a static or blank field). As can be seen from both of these graphs, the MAE is very sensitive toadaptation duration: maximal MAE’s only occur with adapting motion close to the full, unbroken 30 s (‘baseline’). This all but rules out maximal MAE’s coming from anyleakage during withdrawn conditions.

1180 E. Blaser, T. Shepard / Vision Research 49 (2009) 1174–1181

late cortical activity between the auditory and visual systems(Shomstein & Yantis, 2004), and there have been results showingthat MAE duration (Houghton & Macken, 2003) and cortical activ-ity in visual area MT+ are attenuated by attention to an auditorytask (Berman & Colby, 2002)).

Interestingly, there are cases where prima facie ‘competing’ de-mands do not diminish the processing of other stimuli. Feature-based attentional spreading is a good example, where neural re-sponses and adaptation processes can be influenced in regions ofthe visual field that are not willfully attended (Arman et al.,2006; Melcher et al., 2005; Saenz et al., 2002; Serences & Boynton,2007; Treue & Martinez Trujillo, 1999; Treue & Maunsell, 1996).The results of Arman et al. (2006) are especially illustrative here.There observers directed ‘attention’ to a moving field while MAE’swere measured from an secondary adaptation stimulus in an ig-nored region of the display. This stimulus yielded stronger MAE’swhen the target of attention was a similarly moving field, as op-posed to when it was an oppositely moving one. In what seemsat first blush paradoxical, attention could be withdrawn from anadapting stimulus, yet the result was not a reduction in the MAEbut an enhancement. (This again highlights the inadequacy of theterm ‘attention’. One can speak of devoting ‘attention’ to the targetadaptor and withdrawing it from the secondary adaptor, but thisuse includes the manipulation of resources that are apparently epi-phenomenal to the MAE (e.g. awareness) while failing to ade-quately capture the manipulation of actual processing resources.What really determined whether processing resources were de-voted to the secondary adaptor or not was determined by ‘context’:the motion direction of target adaptor.) This result dovetails nicelywith other results, including Chaudhuri’s less discussed secondexperiment (Chaudhuri, 1990), where ‘attention’ was withdrawnto a color discrimination task, but where the to-be-judged colorswere part of the adapting field itself: MAE’s were undiminished.‘Cross-attribute’ (Sohn et al., 2004) spreading ensured that whatwe term dark attention was devoted both to the color and the mo-tion of the moving field.‘

5. Conclusion

The distinction we emphasize here between visual awarenessand underlying visual processing resources owes much to earlierattention models (Folk, 1992; Neisser, 1967; Schneider & Shiffrin,1977; Lavie, 2006; Braun & Julesz, 1998; Pashler, 2001; VanRullen

et al., 2004). Exploiting this distinction helps illuminate the rules ofunsupervised resource allocation (what we call ‘dark attention’),but also if awareness as well as other cognitive processes can beteased away from, say, a motion stimulus without diminishingits power to yield a MAE, then such an arrangement could providefor cleaner behavioral and neurophysiological studies.

References

Alais, D., & Blake, R. (1999). Neural strength of visual attention gauged by motionadaptation. Nature Neuroscience, 2, 1015–1018.

Arman, A., Ciaramitaro, V., & Boynton, G. (2006). Effects of feature-based attentionon the motion aftereffect at remote locations. Vision Research, 46, 2968–2976.

Bahrami, B., Carmel, D., Walsh, V., Rees, G., & Lavie, N. (2008). Unconsciousorientation processing depends on perceptual load. Journal of Vision, 8, 1–10.

Berman, R. A., & Colby, C. L. (2002). Auditory and visual attention modulate motionprocessing in area MT+. Cognitive Brain Research, 14, 64–74.

Blaser, E., Papathomas, T., & Vidnyánszky, Z. (2005). Binding of motion and colour isearly and automatic. European Journal of Neuroscience, 21, 2040–2044.

Blaser, E., Pylyshyn, Z., & Holcombe, A. (2000). Tracking an object through featurespace. Nature, 408, 196–199.

Brand-D’Abrescia, M., & Lavie, N. (2008). Task coordination between and withinsensory modalities: Effects on distraction. Perception & Psychophysics, 70,508–515.

Braun, J., & Julesz, B. (1998). Withdrawing attention at little or no cost: Detectionand discrimination tasks. Perception & Psychophysics, 60, 1–23.

Braun, J., & Sagi, D. (1990). Vision outside the focus of attention. Perception &Psychophysics, 48, 45–58.

Chaudhuri, A. (1990). Modulation of the motion aftereffect by selective attention.Nature, 344, 60–62.

Ciaramitaro, V. M., & Boynton, G (2007). Psychophysical correlates of the effects ofcross-modal attention are more consistent with fMRI responses in V1 than MT+.Perception, 36, ECVP Abstract Supplement.

Connor, C. E., Preddie, D. C., Gallant, J. L., & Van Essen, D. C. (1997). Spatial attentioneffects in macaque area V4. Journal of Neuroscience, 17, 3201–3214.

Dehaene, S., Changeux, J. P., Naccache, L., Sackur, J., & Sergent, C. (2006). Conscious,preconscious, and subliminal processing: a testable taxonomy. Trends inCognitive Science, 10, 204–211.

Duncan, J., Martens, S., & Ward, R. (1997). Restricted attentional capacity within butnot between sensory modalities. Nature, 387, 808–810.

Duncan, J. (1984). Selective attention and the organization of visual information.Journal of Experimental Psychology: General, 113, 501–517.

Folk, C. L., Remington, R. W., & Johnston, J. C. (1992). Involuntary covert orienting iscontingent on attentional control settings. Journal of Experimental Psychology:HPP, 18, 1030–1044.

Georgiades, M., & Harris, J. (2000). The spatial spread of attentional modulation ofthe motion aftereffect. Perception, 29, 1185–1201.

Hardcastle, V. G. (1997). Attention versus consciousness: a distinction with adifference. Cognitive studies. Bulletin of the Japanese Cognitive Science Society, 4,56–66.

He, S., Cavanagh, P., & Intriligator, J. (1996). Attentional resolution and the locus ofvisual awareness. Nature, 383, 334–337.

Houghton, R. J., Macken, W. J., & Jones, D. M. (2003). Attentional modulation of thevisual motion aftereffect has a central cognitive locus: Evidence of interference

E. Blaser, T. Shepard / Vision Research 49 (2009) 1174–1181 1181

by the postcategorical on the precategorical. Journal of Experimental Psychology:HPP, 29, 731–740.

Huk, A. C., Ress, D., & Heeger, D. J. (2001). Neuronal basis of the motion aftereffectreconsidered. Neuron, 32, 161–172.

James, W. (1890). The principles of psychology. New York: Holt (pp. 403–404).Kentridge, R., Heywood, C., & Weiskrantz, L. (2004). Spatial attention speeds

discrimination without awareness in blindsight. Neuropsychologia, 42, 831–835.Koch, C., & Tsuchiya, N. (2007). Attention and consciousness: Two distinct brain

processes. Trends in Cognitive Science, 11, 16–22.Lamme, V. (2003). Why visual attention and awareness are different. Trends in

Cognitive Sciences, 7, 12–18.Lankheet, M., & Verstraten, F. (1995). Attentional modulation of adaptation to two-

component transparent motion. Vision Research, 35, 1401–1412.Lavie, N. (1995). Perceptual load as a necessary condition for selective attention.

Journal of Experimental Psychology: Human Perception and Performance, 21,451–468.

Lavie, N. (2006). The role of perceptual load in visual awareness. Brain Research,1080, 91–100.

Mack, A., & Rock, I. (1998). Inattentional blindness. Cambridge: MIT Press.Mather, G. (1980). The movement aftereffect and a distribution-shift model for

coding the direction of visual movement. Perception, 9, 379–392.Mather, G., & Harris, J. (1998). Theoretical models of the motion aftereffect. The

motion aftereffect: A modern perspective (pp. 157–185). Cambridge: MIT Press.Melcher, D., Papathomas, T. V., & Vidnyanszky, Z. (2005). Implicit attentional

selection of bound visual features. Neuron, 6, 723–729.Merikle, P., & Joordens, S. (1997). Parallels between perception without attention

and perception without awareness. Consciousness and Cognition: AnInternational Journal, 6, 219–236.

Montaser-Kouhsari, L., & Rajimehr, R. (2005). Subliminal attentional modulation incrowding condition. Vision Research, 45, 839–844.

Moray, N. (1959). Attention in dichotic listening: Affective cues and the influence ofinstructions. Quarterly Journal of Experimental Psychology, 11, 56–60.

Naccache, L., Blandin, E., & Dehaene, S. (2002). Unconscious masked primingdepends on temporal attention. Psychological Science, 13, 416–424.

Neisser, U. (1967). Cognitive Psychology. New York: Appleton-Century-Crofts.O’Regan, J., & Noë, A. (2001). A sensorimotor account of vision and visual

consciousness. Behavioral and Brain Sciences, 24, 939–1031.O’Craven, K. M., Downing, P. E., & Kanwisher, N. (1999). fMRI evidence for objects as

the units of attentional selection. Nature, 401, 584–587.Pashler, H., & Johnston, J. (1998). Attentional limitations in dual-task performance.

Attention (pp. 155–189). Hove England (UK) Taylor & Francis: Psychology Press/Erlbaum.

Pashler, H., Johnston, J. C., & Ruthruff, E. (2001). Attention and performance. AnnualReview of Psychology, 52, 629–651.

Posner, M. I. (1994). Attention: The mechanisms of consciousness. Proceedings of theNational Academy of Sciences of the United States of America, 91, 7398–7403.

Rees, G., Frith, C., & Lavie, N. (1997). Modulating irrelevant motion perception byvarying attentional load in an unrelated task. Science, 278, 1616–1619.

Rees, G., Frith, C., & Lavie, N. (2001). Processing of irrelevant visual motion duringperformance of an auditory attention task. Neuropsychologia, 39, 937–949.

Rensink, R. A., O’Reagan, J. K., & Clark, J. J. (1997). To see or not to see: The need forattention to perceive changes in scenes. Psychological Science, 8, 368–373.

Rezec, A., Krekelberg, B., & Dobkins, K. (2004). Attention enhances adaptability:Evidence from motion adaptation experiments. Vision Research, 44, 3035–3044.

Saenz, M., Buracas, G. T., & Boynton, G. M. (2002). Global effects of feature-basedattention in human visual cortex. Nature Neuroscience, 5, 631–632.

Schneider, W., & Shiffrin, R. (1977). Controlled and automatic humaninformation processing: I. Detection, search, and attention. PsychologicalReview, 84, 1–66.

Seiffert, A. E., Somers, D. C., Dale, A. M., & Tootell, R. B. (2003). Functional MRIstudies of human visual motion perception: Texture, luminance, attention andaftereffects. Cerebral Cortex, 13, 340–349.

Serences, J. T., & Boynton, G. M. (2007). Feature-based attentional modulations inthe absence of direct visual stimulation. Neuron, 55, 301–312.

Shomstein, S., & Yantis, S. (2004). Control of attention shifts between vision andaudition in human cortex. Journal of Neuroscience, 24, 10702–10706.

Simons, D. J., & Levin, D. T. (1997). Change blindness. Trends in Cognitive Sciences, 1,261–267.

Sohn, W., Papathomas, T. V., Blaser, E., & Vidnyanszky, Z. (2004). Object-based cross-feature attentional modulation from color to motion. Vision Research, 44,1437–1443.

Sperling, G. (1960). The information available in brief visual presentations.Psychological Monographs, 74 (11, Whole No. 498).

Sperling, G., Budiansky, J., Spivak, J. G., & Johnson, M. C. (1971). Extremely rapidvisual search: The maximum rate of scanning letters for the presence of anumeral. Science, 174, 307–311.

Sperling, G., & Melchner, M. (1978). The attention operating characteristic:Examples from visual search. Science, 202, 315–318.

Sutherland, N. S. (1961). Figural aftereffects and apparent size. Quarterly Journal ofExperimental Psychology, 13, 222–228.

Treisman, A., & Davies, A. (1973). Dividing attention to ear and eye. In S. Kornblum(Ed.), Attention and performance IV (pp. 101–117). New York: Academic Press.

Treisman, A., & Gelade, G. (1980). A feature-integration theory of attention.Cognitive Psychology, 12, 97–136.

Treue, S., & Martinez Trujillo, J. C. (1999). Feature-based attention influences motionprocessing gain in macaque visual cortex. Nature, 399, 575–579.

Treue, S. (2003). Visual attention: the where, what, how and why of saliency.Current Opinion in Neurobiology, 13, 428–432.

Treue, S., & Maunsell, J. H. (1996). Attentional modulation of visual motionprocessing in cortical areas MT and MST. Nature, 382, 539–541.

VanRullen, R., Reddy, L., & Koch, C. (2004). Visual search and dual tasks reveal twodistinct attentional resources. Journal of Cognitive Neuroscience, 16, 4–14.

Velmans, M. (1996). The science of consciousness. Routledge.Verstraten, F. A. J., Fredericksen, R. E., van Wezel, R. J. A., Lankheet, M. J. M., &

van de Grind, W. A. (1996). Recovery from adaptation for dynamic and staticmotion aftereffects: Evidence for two mechanisms. Vision Research, 36,421–424.

von Grünau, M., Bertone, A., & Pakneshan, P. (1998). Attentional selection of motionstates. Spatial Vision, 11, 329–347.

Whitney, D., & Bressler, D. W. (2007). Second-order motion without awareness:Passive adaptation to second-order motion produces a motion aftereffect. VisionResearch, 47, 569–579.

Wohlgemuth, A. (1911). On the after-effect of seen movement. British Journal ofPsychology Monograph. Oxford: University Press.

Wolfe, J. M., Cave, K. R., & Franzel, S. L. (1989). Guided search: An alternative to thefeature integration model for visual search. Journal of Experimental Psychology:Human Perception and Performance, 15, 419–433.