Vision Research 51 (2011) 323–332

Contents lists available at ScienceDirect

Vision Research

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

Location and identity memory of saccade targets

I-Fan Lin, Andrei Gorea ⇑Laboratorie Psychologie de la Perception, Paris Descartes University and CNRS, 45 rue des Saints Pères, 75006 Paris, France

a r t i c l e i n f o a b s t r a c t

Article history:Received 4 August 2010Received in revised form 17 October 2010Available online 27 November 2010

Keywords:Spatial constancySaccadesSpatial memoryReference frameObject-identity memory

0042-6989/$ - see front matter � 2010 Elsevier Ltd. Adoi:10.1016/j.visres.2010.11.010

⇑ Corresponding author.E-mail address: andrei.gorea@parisdescartes.fr (A.

1 In the context of this study ‘‘frame-based’’ and ‘‘spaand will be used interchangeably.

While the memory of objects’ identity and of their spatiotopic location may sustain transsaccadic spatialconstancy, the memory of their retinotopic location may hamper it. Is it then true that saccades perturbretinotopic but not spatiotopic memory? We address this issue by assessing localization performances ofthe last and of the penultimate saccade target in a series of 2–6 saccades. Upon fixation, nine letter-pairs,eight black and one white, were displayed at 3� eccentricity around fixation within a 20� � 20� greyframe, and subjects were instructed to saccade to the white letter-pair; the cycle was then repeated. Iden-tical conditions were run with the eyes maintaining fixation throughout the trial but with the grey framemoving so as to mimic its retinal displacement when the eyes moved. At the end of a trial, subjectsreported the identity and/or the location of the target in either retinotopic (relative to the current fixationdot) or frame-based1 (relative to the grey frame) coordinates. Saccades degraded target’s retinotopic loca-tion memory but not its frame-based location or its identity memory. Results are compatible with thenotion that spatiotopic representation takes over retinotopic representation during eye movements therebycontributing to the stability of the visual world as its retinal projection jumps on our retina from saccade tosaccade.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction efficiency of these processes in stabilizing the visual world has been

Most visual tasks, from viewing natural scenes to reading thisarticle, require sequential inspection of an array of objects via sacc-adic eye movements so that selected visual details can be resolved,identified, stored and retrieved. By so doing, however, saccades en-tail a continuously jumping retinal projection of the world. The factthat we do not experience its permanent jitter has been an endur-ing scientific puzzle (see Burr & Morrone, 2010).

It has been proposed that the perceptual stability of the visualworld is due to a number of jointly active processes such as retinalmotion cancellation by proprioceptive signals and/or by an efferentcopy (Bridgeman, 1995; Dodge, 1990; Helmholtz, 1867; Holst &Mittelstaedt, 1971 (1950); Sperry, 1950; Stark & Bridgman, 1983),saccadic suppression (e.g. Bridgeman, Hendry, & Stark, 1975; Burr,Morrone, & Ross, 1994; Dodge, 1990; Matin, 1974; Volkmann,Schick, & Riggs, 1968), reafferent visual information (Deubel, 2004;Deubel & Schneider, 1994; Deubel, Schneider, & Bridgeman, 1996,2002) including optic flow (Gibson, 1966), remapping (Cavanagh,Hunt, Afraz, & Rolfs, 2010; Colby & Goldberg, 1999; Duhamel,Bremmer, Ben Hamed, & Graf, 1997; Duhamel, Colby, & Goldberg,1992; Wurtz, 2008) and spatiotopic coding (Dodge, 1990; Holst &Mittelstaedt, 1971(1950); Sperry, 1950). It remains that the

ll rights reserved.

Gorea).tiotopic’’ are equivalent terms

consistently challenged (see reviews by Cavanagh et al., 2010;Deubel et al., 2002; Wurtz, 2008).

The focal topic of the present study is spatiotopic (or reference-frame-based) location coding and storage. Spatiotopic locationstorage being independent of eye location should enhance the sta-bility of the perceived world. Retinotopic location storage shouldhinder it. Of course, the brain may store objects’ location in bothreference frames but at some cost (sensory, mnemonic, attentional,etc.). If such cost were to be avoided so as to favor world stabilitywhen the eyes move, preservation of spatiotopic memory shouldprevail over retinal location mnemonic traces. In fact, one mayconjecture that spatiotopic memory might even be enhanced dur-ing eye-movements compared to fixation conditions where retino-topic coordinates suffice to localize objects in the world.

One form of transsaccadic spatiotopic storage is what has beenreferred to as ‘transsaccadic fusion’, namely the correct transsacc-adic visual ‘pasting’ of parts of the same visual object separatelypresented before and after the saccade (e.g. Breitmeyer, Kropfl, &Julesz, 1982; Jonides, Irwin, & Yantis, 1982; McConkie & Rayner,1976). As experimentally defined, transaccadic fusion requires onlyshort-lived memory (the time of a saccade) but high spatial accu-racy that can be only achieved at the sensory level. In other words,transsaccadic fusion, if it exists, is likely to involve iconic ratherthan short term memory. It remains that the very existence oftranssaccadic fusion has been repeatedly contested on methodo-logical (see Irwin, 1996) and theoretical (Rojer & Schwartz, 1990;Simons & Levin, 1997; Yeshurun & Schwartz, 1989) grounds and/

324 I-Fan Lin, A. Gorea / Vision Research 51 (2011) 323–332

or could not be replicated (DiLollo, 1980; Irwin, Yantis, & Jonidas,1983; Irwin, Zacks, & Brown, 1990; McConkie & Currie, 1996;O’Regan and Lévy-Schoen, 1983; Rayner & Pollatsek, 1983).

Transsaccadic fusion is not the only form of spatiotopic storagehelpful in keeping track of where objects are in the world whilemoving the eyes. Relational information (whereby objects presentacross saccades are used as landmarks for post-saccadic localiza-tion updating; Deubel, 2004; Deubel & Schneider, 1994; Deubelet al., 1996, 2002) stored at a coarser spatial scale (Cohen & Ivry,1991; Irwin, 1996) and possibly at a more symbolic level involvingglobal shapes or schematic maps rather than refined stimulusfeatures (Aivar, Hayhoe, Chizk, & Mruczek, 2005; Hayhoe,Shrivastavah, Mruczek, & Pelz, 2003; Hochberg, 1968; Irwin,1996; Rensink, 2000) may still be useful in entertaining a stableperception of the world over durations larger than iconic memory.

A rather drastic alternative to spatiotopic memory as a means ofpreserving spatial constancy is total post-saccadic forgetting ofboth spatio- and retinotopic pre-saccadic information (Horowitz& Wolfe, 1998; Moore & Egeth, 1997; Wolfe, 1998; Wolfe, Klem-pen, & Dahlen, 2000) or, empirically equivalent, the ‘non-represen-tation’ of such information (O’Regan, 1992; O’Regan and Noe,2001; Rensink, 2000). Even though the notion of a perceptualworld exclusively based on an ‘outside memory’ (O’Regan, 1992)is attractive, it has been consistently refuted by studies havingshown that such transsaccadic storage does exist (see above).

The rationale of the present study is based on the notion thatstorage of objects’ spatiotopic (or frame-based) location promotestranssaccadic spatial constancy, while storage of their retinotopiclocation hinders it. If so, spatiotopic location storage should bemore resistant to (or possibly even enhanced by) eye-movementsthan retinotopic location storage. Whether or not this is the caseis the main question asked in the present study. In addition, wealso inquire into the extent to which transsaccadic location storagecompetes with object identity storage, a topic never addressed toour knowledge.

2. Methods

2.1. Stimuli

Participants were seated in a completely dark room with thehead positioned on a chin rest, 63 cm from a 38� � 28� 22WFormac ProNitron 22800 screen with a spatial resolution of 1024by 768 pixels and a 96 Hz refresh rate. Movements of the righteye were measured using an EyeLink 1000 Desktop Mount (SRResearch, Osgoode, Ontario, Canada) with an average spatial reso-lution of 0.25�, and a 1 kHz sampling rate. The experiment wascontrolled by an Apple Dual Intel-Core Xeon computer; manual re-sponses were recorded via a standard keyboard and mouse. Theexperimental software controlling the stimuli display and responserecording was implemented in MATLAB (MathWorks, Natick, Mas-sachusetts, USA), using the Psychophysics (Brainard, 1997; Pelli,1997) and EyeLink toolboxes (Cornelissen, Peters, & Palmer, 2002).

Stimuli were letter-pairs2 (arial font 20, 0.73� � 0.73�) randomlychosen from 20 consonants (Q, W, R, T, P, S, D, F, G, H, J, K, L, Z, X, C, V,B, N, and M) on each trial. One white letter-pair (6.96 cd/m2) servedas the target, and eight black letter-pairs (0.01 cd/m2) served asdistracters. The nine letter-pairs were equally spaced (2.05� of visualangle) 3� away from fixation (a 0.3� radius bull’s eye) and were dis-played within a 20� � 20� grey frame (1.36 cd/m2) partitioned infour equal quadrants by a black cross-like reticle (0.01 cd/m2; linewidth: 0.07�; see Fig. 1). The background outside the grey framewas black (0.01 cd/m2). The size and contrast of the letters were

2 Presentation of single letters entailed 100% letter identity recall whatever theexperimental condition.

chosen based on preliminary trials so as to insure 100% identificationat 3� eccentricity.

2.2. Procedure

Each trial started with the grey square-frame and its reticledisplayed at the center of the screen and with the fixation bull’seye displayed at a random position 1� away from the middle of thecross-reticle (Fig. 1A). Subjects were instructed to fixate the bull’seye black dot. Once the software detected a steady fixation for150 ms, the target (white letter-pair) and the eight distracters (blackletter-pairs) were simultaneously displayed around fixation (Fig. 1Band F). They were turned off together with the fixation dot after100 ms. Subsequently, subjects were presented with one out oftwo main stimulation sequences, a ‘saccade’ or a ‘fixation’ sequence.

In the saccade blocks subjects were instructed to make saccadesto the (white) targets as they appeared at different locations in aseries of 2–6 such changes. In this condition the grey frame andits reticle remained at their central position with respect to thescreen throughout the 2–6 target location (and identity) changesand thus throughout the 2–6 corresponding saccades. For the con-dition where subjects had to report the location or identity of thelast (i.e. 2nd) target (Fig. 1D) there were always two target changes(and hence two saccades). For the condition where subjects re-ported the location or identity of the penultimate (i.e. respectively2nd–5th) target (Fig. 1B) the number of target changes was ran-domized across trials in-between 3 and 6 thus preventing subjectsfrom anticipating the end of a trial. On each saccade landing (as de-tected online; see below), a new fixation dot appeared within oneraster frame at the location of the just extinguished target (namelyclose by the current saccade landing position; Fig. 1C) together withthe new letter-pairs (target and distracters) around and 3� awayfrom it (Fig. 1D). The location of the letter-pairs relatively to the fix-ation dot was always the same. The new fixation dot, target anddistracters were offset after 100 ms and the cycle repeated (Fig. 1illustrates a cycle of two saccades and two frame shifts only).

In the fixation blocks, subjects were instructed to fixate thebull’s eye throughout the trial. The bull’s eye, target and distracterswere also offset 100 ms after their onset, with the grey frame andits reticle remaining at their current position for a duration equalto the average saccade latency (as determined for each subject inpreliminary trials). Following this duration, the grey frame andits reticle were shifted by 3� in a direction opposite to the targetposition relative to the bull’s eye (orange arrow in Fig. 1G) so asto mimic the movement on the retina in the saccade blocks. Thismovement was programmed as a sequence of four jumps over a to-tal duration of 50 ms (the average time of a saccade). Once the greyframe reached its new location, a new set of target and distractersappeared around the immobile fixation dot at their initial positions(with respect to the screen but not to the grey frame; Fig. 1H). As inthe saccade blocks, localization or identity reports of the last targetwere always made in a sequence of two letter-pairs presentations(and two frame shifts), while reports of the penultimate target weremade for presentation sequences (and frame shifts) of 3–6.

At the end of each saccade or fixation trial and whatever the(position or identity) recall task, the fixation dot changed its colorto red for 150 ms, after which the grey frame was shifted (withinone raster frame) to a new pseudo-random location with the sizeand direction of the shift constrained so that one of the nine re-sponse boxes (randomly chosen) appeared at the spatiotopic loca-tion of the relevant (last or penultimate) target in either theretinotopic or spatiotopic location recall blocks (Fig. 1K and L; or-ange and green arrows show the gray frame shifts with respectto its previous position in E and I, respectively; note that as thelocation of the letter-pairs relative to the fixation dot never chan-ged, this proviso was futile for the retinotopic recall task. Given

Fig. 1. Spatiotemporal layout of one trial involving the last two saccades in the Saccade-Fixed frame condition (A–E and J, K or L) or the two frame shifts in the Fixation-Moving frame conditions (A, F–I and J, K or L) trial. (In the actual ‘penultimate target’ experiments the number of saccades or frame shifts varied randomly in-between 3 and6.) In both conditions, a trial started with an bull’s eye fixation dot (A) followed by a 100 ms presentation of eight black (distracters) and of one white (target) letter-pairsdisplayed on an invisible (shown here for illustration purposes), 3� radius circle around fixation (B and F). In the ‘saccade’ condition, observers had to saccade to the white(target) letter-pair (white arrows in B and D; in this illustration B shows the penultimate target and saccade) which was replaced upon detection of the saccade landing by anew fixation dot (C) followed within one raster frame by a new set of distracters and target (D; last target and saccade); in the ‘fixation’ condition, saccades were replaced byequivalent shifts of the grey frame indicated by the orange arrows in G and H with the fixation dot (G) and letter-pairs around it (H) appearing after intervals matched to thesaccade condition. Note that in the absence of eye-movements the fixation dot and letter-pairs did not change position with respect to the screen (black rectangle) throughoutthe whole trial. At the end of the saccade or frame shift sequence, the fixation dot turned red (E and I) and, depending on the task, subjects were presented within the nextraster frame with (i) either nine grey boxes around their current fixation only one of which occupied the retinal (blue boxes) or spatiotopic location (yellow boxes) of the last(K) or penultimate (L) target (localization task), or (ii) 18 consonants displayed on two rows of nine each including one letter of the target letter-pair (J; identification task). Inall three cases (J–L), the grey frame underwent a final shift whose size was constrained so that one, randomly chosen response boxe be at the correct spatiotopic location(with equivalent shifts used in both retinotopic and identification recall tasks). The green and orange arrows in K and L indicate the grey frame shifts relative to the last frameof the ‘saccade’ (E) and ‘fixation’ (I) conditions, respectively. The spatial layouts are not at scale.

3 Note that the forced choice response format used here for all tasks allows thedirect comparison of the corresponding performances in d0 units. At least, this shouldbe the case according to Signal Detection Theory, that provides transformation toolsfrom % correct to d0 whatever the number of forced choice alternatives. Such directcomparisions were not possible in previous studies most of which measuredlocalization dispersion and percent correct identification.

I-Fan Lin, A. Gorea / Vision Research 51 (2011) 323–332 325

that spatiotopic and retinotopic locations are mirror symmetricwith respect to fixation, this final grey frame shift was meant toprevent observers from inferring one from the other in the last tar-get recall blocks. For homogeneity, final frame shifts were appliedto all other experimental conditions.

In all experimental blocks involving a localization task, theshift of the grey frame was immediately (one raster frame) fol-lowed by the presentation of nine grey squares (0.88� � 0.88�)displayed around the current fixation with the same layout asthe letter-pairs. Only one of these squares coincided with target’sretinotopic or spatiotopic position (blue and yellow boxes, respec-tively, in Fig. 1K and L). Subjects made their (retinotopic or spa-tiotopic) localization response by clicking on one of the nineboxes. The retinotopic position of a target was defined with re-spect to the fixation dot at the moment when this target was pre-sented; at the end of a trial subjects had to retrieve it withrespect to the current fixation dot. The spatiotopic position of atarget was defined with respect to the grey frame (and its yokedreticle) at the moment when this target was presented; at the end

of a trial subjects had to retrieve it with respect to the currentlocation of the grey frame.

In the target identification blocks, 18 out of the 20 possible con-sonants were displayed in two rows of nine different letters aboveand below fixation (Fig. 1J). The letters and their order werepseudo-randomized across trials so that the top and bottom rowscontained respectively the first and second letter of the targetletter-pair. Subjects were told so and had to click on these twoletters. Separate blocks were run where subjects had to report ona random basis either the location (50%) or the identity of thetarget (referred hereafter as the dual-task). The response-typewas revealed at the end of each trial by the display of either thelocalization squares or the two rows of letters.3 Response times

326 I-Fan Lin, A. Gorea / Vision Research 51 (2011) 323–332

(the time interval between the onset of the ‘location squares’ or ofthe letters and subject’s mouse click) were also recorded (eventhough subjects received no response speeding instruction).

There were altogether 20 experimental conditions as defined by:(1) the target to be memorized (last or penultimate; hereafterreferred to as distinct experiments), (2) eye-movement behavior(saccade or fixate), (3) task type specified by the memorizationinstructions (single-tasks: retinotopic or spatiotopic localizationor target identity; dual-tasks: retinotopic localization and targetidentity or spatiotopic localization and target identity) (see Table 1).

At the beginning of each ‘last’ or ‘penultimate’ target block, sub-jects received one of five instructions, namely to memorize (last orpenultimate) target’s retinotopic or spatiotopic location, or itsidentity, or both target’s retinotopic location and identity, or tar-get’s spatiotopic location and identity.

All subjects got familiarized with each of the 20 different exper-imental conditions during 4–5 training hours. One subject (out ofseven) was discarded based on her close to chance performancein a number of conditions after 5 h of training. Each ‘single-task-last-target’ condition consisted of 96 trials run in two blocks of48 trials (see Table 1). Each ‘dual-task-last-target’ condition con-sisted of 192 trials run in four blocks of 48 trials. Out of the total28 last-target blocks, 14 (7 ‘saccade’ and 7 ‘fixation’) were run atthe mid-term of the whole experimental period and the other 14at the end of this period. The order was randomized across subjects.Subjects ran 120 trials for each ‘single-task-penultimate-target’condition (in 5 blocks of 24 trials) and 240 trials for each ‘dual-task-penultimate-target’ condition (in 10 blocks of 24 trials). The70 penultimate-target blocks were split into 10 sessions of 7 blocks(of 24 trials) with the order of the single and dual conditionsrandomized across sessions and observers. Subjects were run insessions of 1–3 h a day with breaks every half hour or whenever theyfelt tired for a total of 9–12 days distributed over two to 3 weeks.

2.3. Subjects

Six participants (three females; age range 21–38) completed thefull experimental set. They all had normal or corrected-to-normalvision and gave their informed consent. The experiments were car-ried out according to the ethical standards laid down in the Decla-ration of Helsinki.

3. Results

3.1. Data analysis

Eye movements were analyzed online to assess fixations, sac-cade onsets and landing positions. The eyes were said to fixate aslong as the position of the tracked eye was within a radius of1.5� about the fixation dot for at least 150 ms. A saccade was saidto occur (and its trajectory was tracked to the next landing posi-tion) when eye’s velocity exceeded 30�/s. A saccade was consideredto come to an end once eye velocity dropped below 30�/s. Trialswere discarded if saccade latency exceeded 500 ms or if subjectscould not stabilize their eye position within the 1.5�-window over500 ms. In the ‘fixation’ blocks, trials were discarded if observer’seyes left the 1.5�-window around the fixation dot before the greyframe shift. Discarded trials were reinserted at the end of a block.

Both localization and identity performance is given in d0 units. d0

values were derived from the percent correct (see Macmillan &Creelman, 2005) obtained in the 9AFC localization trials and inthe 81AFC ‘identity’ trials4 for each experimental condition and

4 Observers had to choose two letters, each one out of a different group of nineletters. Hence the number of possible choices is 81.

subject. Given that a d0 derivation from a 81AFC procedure may lookawkward, percent correct identification performances have alsobeen transformed into their arcsine and processed as such in thestatistical analysis. Response times (RTs) from the onset of theresponse boxes or letters to subject’s click on one of them were alsomeasured and processed statistically.

3.2. Sensitivity (d0)

Fig. 2 displays localization (left panels) and identification (rightpanels) performances for the last (top) and penultimate (bottom)target experiments under fixation and saccade conditions. Solidand open symbols are for single- (localization or identification)and dual-tasks (localization and identification). Circles and trian-gles denote retinotopic and frame-based localization perfor-mances, or identification performances in the correspondingdual-tasks. Saccades appear to deteriorate retinotopic localizationperformance (by about 0.3 d0-units) and enhance by about as much(although this enhancement fails to reach statistical significance;see below) frame-based localization performance be they assessedin single- or dual-task conditions. Instead, saccades do not seem toaffect identification performances whether in the single- or dual-task. Dual- (compared to single-) task conditions appear to deteri-orate mostly spatiotopic localization (but the statistical analysisbelow also shows a significant drop in the identification of the pen-ultimate target). Finally, both localization and identification of thelast target are �1.2 d0-units better than of the penultimate target.This is understandable given memory fading (see Bays & Husain,2008) and because the respective performances have been ob-tained under critically different conditions (fixed vs. randomizednumber of targets per trial; see Section 2).

Two three-way ANOVAs for the localization data (factors: eye-movements, EM (fixate, saccade), reference frame, RF (retinotopic,frame-based), task type, TT (single, dual)) and two two-way ANO-VAs for the identification data (factors EM and TT) separately per-formed for the ‘last’ and ‘penultimate target’ experiments5 confirmmost of the observations above. Identification performances wereexpressed in both d0 and arcsin (%-correct) units with equivalent sta-tistical outcomes; only the d0 analysis is given below. (a1) Last sac-cades – localization: only the TT factor shows a significant effect(F(1, 5) = 9.83, p = 0.0258) with the RF factor close to significance(F(1, 5) = 3.84, p = 0.10). Partial comparisons show that the dual-taskaffects spatiotopic localization but only marginally retinotopiclocalization (spatiotopic: F(1, 5) = 9.94, p = 0.0253; retinotopic:F(1, 5) = 4.75, p = 0.0811). The interaction between factors EM andRF is not statistically significant (F(1, 5) = 3.27, p = 0.130), althougha partial comparison shows a marginal EM effect for the retinotopiclocalization (F(1, 5) = 4.27, p = 0.0938). (a2) Last saccades – identifica-tion: the two-way ANOVA shows no effect of either of the two factorsand no significant interaction (all p values > 0.18). (a3) Penultimatesaccades – localization: here again the only significant factor is TT(F(1, 5) = 9.20, p = 0.029), with the RF factor coming close to signifi-cance (F(1, 5) = 5.25, p = 0.07). This time both EM � TT(F(1, 5) = 51.65, p = 0.0008) and EM � RF (F(1, 5) = 9.1, p = 0.0295)interactions are highly significant. The EM � TT interaction is mostlydue to the fact that the dual-task entails a significant spatiotopic(F(1, 5) = 37.87, p = 0.0016) but not retinotopic localization perfor-mance drop. The EM � RF interaction is mostly due to the fact thatsaccades entail a significant drop of the retinotopic localizationperformance (F(1, 5) = 28.79, p = 0.003) with the observed rise inthe spatiotopic localization performance failing to reach significance

5 There are least three ways in which these different experimental conditionsdiffered thereby justifying separate statistical analyses. These reasons are detailed inSection 4.

Table 1All experimental conditions with number of trials per condition (bold digits) and per block (with number of blocks per condition given in italics).

Last target Penultimate target

Saccade Fixation Saccade Fixation

No. of saccades or of gray frame shifts per trial h 2 2 3–6 3–6

Memory tasksSingle Retinotopic location of target 96 (48 � 2) 96 (48 � 2) 120 (24 � 5) 120 (24 � 5)

Spatiotopic location of target 96 (48 � 2) 96 (48 � 2) 120 (24 � 5) 120 (24 � 5)Identity of target 96 (48 � 2) 96 (48 � 2) 120 (24 � 5) 120 (24 � 5)

Dual Retinotopic location + identity of target 192 (48 � 4) 192 (48 � 4) 240 (24 � 10) 240 (24 � 10)Spatiotopic location + identity of target 192 (48 � 4) 192 (48 � 4) 240 (24 � 10) 240 (24 � 10)

a b

dc

Fig. 2. Mean localization (a and c) and identification (b and d) sensitivity (d0) measurements for the ‘last’ (a and b) and ‘penultimate target’ experiments. Filled and opensymbols are for single and dual-tasks, respectively. Circles and triangles are for conditions involving retinotopic and spatiotopic localizations, respectively; squares are for thesingle identification task. Bars are ±1SE.

I-Fan Lin, A. Gorea / Vision Research 51 (2011) 323–332 327

(p = 0.17). (a4) Penultimate saccades – identification: TT is theonly factor yielding a close to significant effect (F(2, 10) = 3.94,p = 0.054).

3.3. Reaction times (RTs)

Even though observers were not asked to perform the localiza-tion or identification tasks in a speeded mode (with all emphasisput on response accuracy), their RTs remain informative aboutthe relationship between retinotopic and frame-based localizationcoding/retrieving. The sensitivity drop and enhancement observedin the saccade (relative to the fixation) conditions for the retino-topic and spatiotopic localizations, respectively, might simply re-flect the well known speed-accuracy trade-off effect (e.g. Palmer,Huk, & Shadlen, 2005). If so, the pattern of RTs as a function of eyes’

behavior (fixate or saccade) and of the localization type (retinotop-ic or spatiotopic) should be the same as the pattern observed forthe corresponding sensitivity measurements (i.e. d0 and RT shouldcorrelate positively). Instead, Fig. 3 (same conventions as Fig. 2)shows that median RTs correlate negatively with d0 (see Fig. 2) forall three dimensions studied (i.e. eye-movements, reference frameand task type) in the localization task but only for the task typedimension in the identification task. As for the sensitivity data,the dual-task yields globally worse performances (i.e. longer RTs)than the single-task even though such drop appears to be negligi-ble for the localization task in the ‘penultimate target’ experiment.Once again, the largest dual performance drop (i.e. RT increase) isobserved for conditions involving the spatiotopic localization task.That observers are globally slower in the spatiotopic than in theretinotopic localization task (by up to 220 ms) suggests that the

a b

dc

Fig. 3. Median localization (a and c) and identification (b and d) RTs for the last (a and b) and penultimate target conditions. Same notations as in Fig. 2. Bars are ±1SE.

328 I-Fan Lin, A. Gorea / Vision Research 51 (2011) 323–332

former, at least in the present experimental format, involves anadditional retrieval process possibly checking spatial relationships(e.g. Deubel, Wolf, & Hauske, 1984; Irwin, McConkie, Carlson-Radvansky, & Currie, 1994).

Once again, two three-way ANOVAs for the localization data(factors EM, RF and TT) and two two-way ANOVAs for the identifi-cation data (factors EM and TT) separately performed for the ‘last’and ‘penultimate target’ experiments (see Note 5 and Section 4)support most of the qualitative observations above. (a1) Lastsaccades – localization: only the TT factor shows a significant effect(F(1, 5) = 16.33, p = 0.0099). Of critical interest, there is also a ten-dency for factors EM and RF to interact (F(1, 5) = 4.09, p = 0.09).(a2) Last saccades – identification: as above, only the TT factorshows a significant effect (F(1, 5) = 15.31, p = 0.0009) with signifi-cantly longer RTs for the identification coupled with spatiotopicthan with retinotopic localization (spatiotopic: F(1, 5) = 27.96,p = 0.0061; retinotopic: F(1, 5) = 5.06, p = 0.0742). (a3) Penultimatesaccades – localization: none of the three factors shows a significanteffect but, as for the d0 data above, the EM � RF interaction is sig-nificant (F(1, 5) = 7.54, p = 0.0405). Also as for the d0 data, this inter-action is mostly due to the fact that saccades entail a significantdrop of the retinotopic localization performance (F(1, 5) = 18.00,p = 0.0081) with the visible mean spatiotopic localization RT short-ening failing to reach significance. (a4) Penultimate saccades – iden-tification: once again, only the TT factor yields a significant effect(F(2, 10) = 6.81, p = 0.0136) with significant differences betweensingle and dual-spatiotopic RTs (F(1, 5) = 7.0, p = 0.0456) but notbetween single and dual-retinotopic RTs.

3.4. Distribution of the localization responses

Figs. 4 and 5 show histograms of observers’ retinotopic and spa-tiotopic localization choices (out of the nine equally spaced loca-tions at 3� eccentricity around fixation), respectively. The angulardifference from the correct location was computed as the differ-ence between the actual and correct answer azimuth, with the fix-ation dot as a reference. Correct choices are shown at 0�. White andblack bars are for fixation and saccade conditions, respectively. Topand bottom panels are for the ‘last’ and ‘penultimate target’ exper-iments, respectively. Data for single and dual conditions are shownin the left- and right-hand panels. The general observation is thatthe two locations closest to the target are reported slightly morefrequently than those far away from it and that, in the ‘penultimatetarget’ experiment, the two locations on the response circle oppo-site to the target location (i.e. ±160� away from it on the responsecircle) tend to be chosen slightly more frequently than the inter-mediate locations. This being said, the distribution of the retino-topic localizations errors is pretty flat. The implication is thatwhen observers do not remember the location of the target theymostly guess. This should be expected given that two adjacentlocations in the present design were separated by 2.05� of visualangle, a distance well beyond the uncertainty range of a transsacc-adic spatial memory evidenced in previous studies (less than 1.4�for saccades within a range of 6–10�; e.g. Collins, Rolfs, Deubel, &Cavanagh, 2009; Deubel, 2004; Karn, Moeller, & Hayhoe, 1997).

Two 3-way ANOVAs (factors: eye-movements, reference frameand task type) separately performed on the standard deviations

-200 -100 0 100 2000

0.5

1

-200 -100 0 100 2000

0.5

1

-200 -100 0 100 2000

0.5

1

-200 -100 0 100 2000

0.5

1sacfix

Fig. 4. Distribution of retinotopic localization responses relative to the correctlocation (0�). White and black bars are for fixation and saccade conditions. Top andbottom panels are for the ‘last’ and ‘penultimate’ target experiments, and left andright panels are for single and dual-task conditions. Vertical lines are ±1SE.

-200 -100 0 100 2000

0.5

1

-200 -100 0 100 2000

0.5

1

-200 -100 0 100 2000

0.5

1

-200 -100 0 100 2000

0.5

1sacfix

Fig. 5. Distribution of spatiotopic localization responses relative to the correctlocation (0�). All conventions are as for Fig. 4.

I-Fan Lin, A. Gorea / Vision Research 51 (2011) 323–332 329

of the localization distributions in the ‘last’ and ‘penultimate tar-get’ experiments show a significant (F(1, 5) = 10.37, p = 0.023)and close to significant (F(1, 5) = 4.83, p = 0.079) task type (singlevs. dual) effect for the ‘last’ and ‘penultimate target’ experiments,respectively. The remaining two factors yield p-values in-between0.17 and 0.96. However, as expected from the corresponding anal-ysis of the sensitivity data, the present ANOVA yields significantTT � RF (F(1, 5) = 23.49, p = 0.005) and EM � RF (F(1, 5) = 9.42,p = 0.028) interactions, comparable to those observed for the sensi-tivity data.

4. Discussion

The present study yields five general observations (i) saccades(compared to fixation conditions) deteriorate retinotopic memoryand possibly enhance spatiotopic memory (even though this lattereffect fails to reach statistical significance); (ii) a dual- (comparedto single-) localization/identification task mostly deteriorates spa-tiotopic localization and to some extent identification perfor-mances; (iii) response times correlate negatively with localizationperformances hence excluding the possibility that the modulationof the latter by eye-movements and reference frame is accountablein terms of a speed-accuracy trade-off; (iv) response times areglobally slower (by up to 220 ms) in the spatiotopic than in theretinotopic localization task suggesting that the former involvesan additional retrieval process; and finally (v) retinotopic localiza-tion errors are close to evenly distributed about the correct loca-tion (with the exception of the two closest locations).

The present research undoubtedly supports the existence of atransaccadic retinotopic and spatiotopic memory trace (see re-views by Cavanagh et al., 2010; Deubel et al., 2002; Wurtz,2008). Hence, along with other studies (e.g. Dickinson & Zelinsky,2007; McCarley, Wang, Kramer, Irwin, & Peterson, 2003; Peterson,Kramer, Wang, Irwin, & McCarley, 2001) including those on inhibi-tion of return (Klein, 2000; Klein & MacInnes, 1999; Müller & vonMühlenen, 2000; Takeda & Yagi, 2000), the present data do notsupport previous claims of an amnesic visual search process(Gilchrist & Harvey, 2000; Horowitz & Wolfe, 1998; Wolfe et al.,2000), nor do they support the notion of a perceptual world exclu-sively based on an ‘‘outside memory’’ (i.e. no ‘‘internal representa-tion’’; O’Regan, 1992). It should be noted, however, that the issue ofa memoryless visual search remains contentious as results dependon a number of factors not always equated across such studies(covert vs. overt attention, persistence or removal of the searcheditems, etc.; for a recent overview see Dickinson & Zelinsky, 2007).Also note that transsaccadic memory needs not and most likelydoes not exhibit the spatial acuity required for transsaccadic fusion(see reviews by Cavanagh et al., 2010; Irwin, 1996; Rensink, 2000).The present technique where localization performance was as-sessed in terms of percent correct with a 9AFC method (nineequally spaced locations on a 3� radius circle) does not allow anaccurate specification of localization accuracy in degrees of visualangle (see below).

The main present finding is the pernicious effect entailed bysaccades on retinotopic but not on spatiotopic localization perfor-mances. While this effect is about equal in size for last andpenultimate saccade target reports, it reaches statistical signifi-cance only for the latter (see below). Compared to conditionswhere the eyes do not move (but the frame of reference does),saccadic eye movements (with a physically stable frame of refer-ence) worsen retinotopic localization performances (a sensitivitydrop of about 0.3 d0 units and a response time increase of about100 ms) but do not affect, or possibly even improve spatiotopiclocalization performances even though this trend is not statisti-cally significant. Such relative improvement (if confirmed in futureexperiments) would ensue from the fact that acting upon objects inspace requires their spatiotopic coding when the eyes move butcan be successfully accomplished based on their retinotopic loca-tion with an immobile eye.

Many previous experimental designs were such that they couldnot differentiate between eye-centered and frame-based localiza-tion performances. In such designs localization and/or identifica-tion performances may (e.g. Gersch, Kowler, Schnitzer, & Dosher,2009; Irwin, 1992; Lawrence, Myerson, & Abrams, 2004) or maynot (e.g. Bays & Husain, 2008) depend significantly on whetherthe eyes move or not. In the present case no such difference is

6 This upper bound is well beyond the typical accuracy range estimated with directmethods (e.g. Collins et al., 2009; Deubel, 2004; Karn et al., 1997) as it wasconstrained by the sparse layout of the nine location alternatives used in the presentstudy. In other words, the present method is such that the correct responses itmeasures include all mislocalizations within the typical spatial uncertainty range.

330 I-Fan Lin, A. Gorea / Vision Research 51 (2011) 323–332

observed when localization performances are averaged across theeye- and frame-based tasks (see panels a, c of Figs. 2 and 3) onthe assumption that observers give about equal weights to thetwo reference frames (e.g. McGuire & Sabes, 2009; Wong & Mack,1981).

Even though, on visual inspection the eye-movement/referenceframe interaction looks pretty much the same for reports of thepenultimate and last target reports (compare Fig. 2a with c andFig. 3 a with c), it does not reach statistical significance for the lat-ter. There are at least three reasons why the comparison betweenthe two experiments may not be warranted. First, the ‘last target’experiment consisted in only two saccades so that observers hadno uncertainty as to the target whose position and/or identitywas to be reported. As a consequence, observers had to store onlyone target per trial with its memory trace unaffected by storingprior or subsequent targets. This was not the case for the ‘penulti-mate target’ experiment where the rank order of this target in aseries of 3-to-6 successive targets was randomized so that observ-ers had to store two successive targets refreshed up to four timesper trial. Understandably, last target performances are up to about1.2 d0 units higher than penultimate target performances with thecorresponding response times up to about 200 ms shorter. The ab-sence of a statistically significant eye-movement/reference framesensitivity interaction for the ‘last target’ experiment may thus bedue, amongst others, to a sensitivity ceiling effect (as the reliabilityof d0 values larger than 3–3.5 is poor). Second, it is nowadays admit-ted that saccade programming is yoked with an attentional process-ing enhancement at the location of the saccade target (e.g. Kowler,Anderson, Dosher, & Blaser, 1995; Schneider & Deubel, 1995). Be-cause saccades are directed toward the last target and away fromthe penultimate one, the memory of the former should be enhancedand of the latter weakened. Finally, inasmuch as saccade program-ming is in retinotopic coordinates (but see Findlay & Walker, 1999;McGuire & Sabes, 2009; Wurtz, 2008), their conversion into spatio-topic ones may not be needed or performed if no further saccade isto be programmed (e.g. Horaguchi & Sugino, 2008). This would bethe case for the last but not for the penultimate saccade in a series.In short, the memory of a last target has a special status that makesit difficult to compare with any other target. Bays and Husain(2008) data show indeed a massive drop in recalling the penulti-mate compared to the last visited target with a much more attenu-ated recall drop for more time remote visited targets.

A direct comparison between retinotopic and spatiotopic globalperformances also raises issues of debate as they strongly dependon the experimental design. In the present experiments, saccadetargets were always 3� from fixation so that storage of the saccadedirection (but not amplitude) was sufficient for retrieving target’sretinotopic but not frame-based coordinates (as the distance be-tween target and reference frame varied across saccades). Theuse of the nine response-squares occupying the same relative posi-tions as the target and distracter letter-pairs (see Fig. 1) may havealso significantly boosted retinotopic localization performances.This is so because the response-squares at the non-target locationscould have been used as spatial landmarks (reference-object the-ory; Deubel, 2004) for the retinotopic but not for the frame-basedlocation retrieval. Along this same line of argument, an enrichmentof the reference frame with additional spatial landmarks may haveboosted the spatiotopic but not the retinotopic localization perfor-mances. In short, differences between retinotopic and spatiotopiclocalization performances are stimulus and task-design dependentand have no specific meaning when considered by themselves. Thecritical observation, however, is that whatever the retinotopicand spatiotopic performances in the absence of eye-movements,saccades worsen retinotopic performances by, in average, almost1 d0-unit and more than 200 ms response time relatively to spatio-topic performance (penultimate target experiments).

The pernicious saccade effect on retinotopic but not on spatio-topic localization memory should contribute to stabilizing the vi-sual world. This does not exclude the possibility that thedeterioration by saccades of retinotopic location memory couldsimply be a byproduct of the tight link between saccade program-ming and spatial working memory (Pearson & Sahraie, 2003;Theeuwes, Belopolsky, & Olivers, 2009; Wurtz, 2008). Inasmuchas saccades are programmed in retinal coordinates that need tobe continuously refreshed, so should be the memory of theirprevious retinal coordinates (e.g. Lawrence et al., 2004; Liu, Yttri,& Snyder, 2010). Instead, transsaccadic preservation of spatiotopicmemory may reflect the sustained activity of ‘‘gain field’’ (or ‘‘realposition’’) neurons in a number of extrastriate areas (see reviewsby Andersen, 1989; Galletti & Fattori, 2002). Within such a frame-work, the fact that a measurable amount of retinotopic memorysurvives saccades by a considerable amount of time could beattributed to a lingering attentional trace (see Golomb, Pulido,Albrecht, Chun, & Mazer, 2010).

Even though saccades affect differently retinotopic and spatio-topic overall localization performances, no such difference is ob-served between the respective distributions of localization errors.Their analysis (see Figs. 4 and 5) shows that, independently ofwhether subjects perform a retinotopic or a frame-based memorytask and of whether or not they move their eyes, most localizationerrors occur for the two locations flanking the correct one andopposite to it (with respect to fixation). Overall, however, the dis-tribution of localization errors is pretty flat implying that observ-ers’ location storage/retrieval uncertainty is at most ±2.05� (thevisual angle separating two adjacent locations in the present de-sign) for either the retinotopic or frame-based task whether theeyes move or not.6

Unlike location memory, identity memory is independent ofwhether or not the eyes move (see also Lawrence et al., 2004;Bay & Hussein, 2008). One previous study that showed otherwise(but only for items far away from the saccade target) was designedso that identity and location could not be disentangled (e.g. Irwin,1992). It may thus be that previously observed identity memorydepletion was mainly due to localization errors. That object’s loca-tion and identity are separately processed (coded and/or storedand retrieved) is only partly sustained by the present data. Whilein most cases (with the exception of localization sensitivity forthe last target and of the response times for the penultimate target)the statistical analyses show significant identity and location per-formance depletion in the dual- (compared to the single-) task,they also indicate that this depletion is observed for the spatiotopicbut not retinotopic localization-identification coupling. Given theadditional observation that identity and spatiotopic memory areeye-movement independent while retinotopic memory is not,one may speculate that identity and retinotopic location are jointlycoded so that they do not compete for attentional resources (henceno penalty when coupled in a dual-task) but are stored separately(hence differently affected by saccades; e.g. Hollingworth &Rasmussen, in press; Kahneman, Treisman, & Gibbs, 1992;Pylyshyn, 1989; Treisman & Gormican, 1988; Treisman & Sato,1990; Wheeler & Treisman, 2002; Xu & Chun, 2006). Instead,spatiotopic location and object identity might be differently codedand stored (e.g. Curtis, 2006; Ester, Serences, & Awh, 2009;Ghorashi, Enns, Klein, & Di Lollo, 2010; Harrison & Tong, 2009;Todd & Marois, 2004; Umeno & Goldberg, 2001) so that they wouldcompete for attentional resources at the coding stage hence

I-Fan Lin, A. Gorea / Vision Research 51 (2011) 323–332 331

entailing storage/retrieval depletion. Be it as it may, the relation-ship between location and identity coding and storage has been amatter of debate for almost three decades (see Treisman &Gormican, 1988) and remains an unsolved research topic (seeCavanagh et al., 2010).

Acknowledgments

This work was supported by Grant ANR-06-NEURO-042-01 to A.Gorea. We thank Patrick Cavanagh, Martin Rolfs, Mark Wexler andThérèse Collins for helpful discussions.

References

Aivar, M. P., Hayhoe, M. M., Chizk, C. L., & Mruczek, R. E. B. (2005). Spatial memoryand saccadic targeting in a natural task. Journal of Vision, 5(3), 177–193.

Andersen, R. A. (1989). Visual and eye movement functions of the posterior parietalcortex. Annual Review in Neuroscience, 12, 377–403.

Bays, P., & Husain, M. (2008). Dynamic shifts of limited working memory resourcesin human vision. Science, 321, 851–854.

Brainard, D. H. (1997). The psychophysics toolbox. Spatial Vision, 10(4), 433–436.Breitmeyer, B. G., Kropfl, W., & Julesz, B. (1982). The existence and role of

retinotopic and spatiotopic forms of visual persistence. Acta Psychologica, 52,175–196.

Bridgeman, B. (1995). A review of the role of efference copy in sensory andoculomotor control systems. Annals of Biomedical Engineering, 23, 409–422.

Bridgeman, G., Hendry, D., & Stark, L. (1975). Failure to detect displacement of visualworld during saccadic eye movements. Vision Research, 15, 719–722.

Burr, D. C., & Morrone, M. C. (2010). Vision: Keeping the world still when the eyesmove. Current Biology, 20(10), R442–R444.

Burr, D. C., Morrone, M. C., & Ross, J. (1994). Selective suppression of themagnocellular visual pathway during saccadic eye movements. Nature, 371,511–513.

Cavanagh, P., Hunt, A. R., Afraz, A., & Rolfs, M. (2010). Visual stability based onremapping of attention pointers. Trends in Cognitive Sciences, 14, 147–153.

Cohen, A., & Ivry, R. B. (1991). Density effects in conjunction search: Evidence forcoarse location mechanism of feature integration. Journal of ExperimentalPsychology: Human Perception and Performance, 17, 891–901.

Colby, C. L., & Goldberg, M. E. (1999). Space and attention in parietal cortex. AnnualReview in Neuroscience, 22, 319–349.

Collins, T., Rolfs, M., Deubel, H., & Cavanagh, P. (2009). Post-saccadic locationjudgments reveal remapping of saccade targets to non-foveal locations. Journalof Vision, 9(5), 1–9.

Cornelissen, F. W., Peters, E. M., & Palmer, J. (2002). The eyelink toolbox: Eyetracking with MATLAB and the psychophysics toolbox. Behavior ResearchMethods, Instruments, & Computers, 34(4), 613–617.

Curtis, C. E. (2006). Prefrontal and parietal contributions to spatial workingmemory. Neuroscience, 139, 173–180.

Deubel, H. (2004). Localization of targets across saccades: Role of landmark objects.Visual Cognition, 11, 173–202.

Deubel, H., & Schneider, W. X. (1994). Can man bridge a gap? Behavioral and BrainSciences, 17, 259–260.

Deubel, H., Schneider, W. X., & Bridgeman, B. (1996). Post-saccadic target blankingprevents saccadic suppression of image displacement. Vision Research, 36,1827–1837.

Deubel, H., Schneider, W. X., & Bridgeman, B. (2002). Transsaccadic memory ofposition and form. In J. Hyönä, D. P. Munoz, W. Heide, & R. Radach (Eds.), Thebrain’s eye: Neurobiological and clinical aspects of oculomotor research. Progress inbrain research (Vol. 140, pp. 165–180). Amsterdam: Elsevier Science.

Deubel, H., Wolf, W., & Hauske, G. (1984). The evaluation of the oculomotor errorsignal. In A. G. Gale & F. Johnson (Eds.), Theoretical and applied aspects of eyemovement research (pp. 55–62). Amsterdam: Elsevier–North-Holland.

Dickinson, C. A., & Zelinsky, G. J. (2007). Memory for the search path: Evidence for ahigh-capacity representation of search history. Vision Research, 47(13),1745–1755.

DiLollo, V. (1980). Temporal integration in visual memory. Journal of ExperimentalPsychology: General, 109, 75–97.

Dodge, R. (1990). Visual perception during eye movement. Psychological Review, 7,454–465.

Duhamel, J.-R., Bremmer, F., Ben Hamed, S., & Graf, W. (1997). Spatial invariance ofvisual receptive fields in parietal cortex neurons. Nature, 389, 845–848.

Duhamel, J.-R., Colby, C. L., & Goldberg, M. E. (1992). The updating of therepresentation of visual space in parietal cortex by intended eye movements.Science, 255, 90–92.

Ester, E. F., Serences, J. T., & Awh, E. (2009). Spatially global representations inhuman primary visual cortex during working memory maintenance. Journal ofNeuroscience, 29(48), 15258–15265.

Findlay, J. M., & Walker, R. (1999). A model of saccade generation based on parallelprocessing and competitive inhibition. Behavioral and Brain Sciences, 22,661–721.

Galletti, C., & Fattori, P. (2002). Posterior parietal networks coding visual space. InH.-O. M. Karnath & A. D. G. Vallar (Eds.), The cognitive and neural bases of spatialneglect (pp. 59–69). Oxford, UK: Oxford University Press.

Gersch, T. M., Kowler, E., Schnitzer, B. S., & Dosher, B. A. (2009). Attention duringsequences of saccades along marked and memorized paths. Vision Research,49(10), 1256–1266.

Ghorashi, S., Enns, J. T., Klein, R. M., & Di Lollo, V. (2010). Spatial selection and targetidentification are separable processes in visual search. Journal of Vision, 10(3),1–12.

Gibson, J. J. (1966). The senses considered as perceptual systems. Boston: HoughtonMifflin.

Gilchrist, I. D., & Harvey, M. (2000). Refixation frequency and memory mechanismsin visual search. Current Biology, 10(19), 1209–1212.

Golomb, J. D., Pulido, V. Z., Albrecht, A. R., Chun, M. M., & Mazer, J. A. (2010).Robustness of the retinotopic attentional trace after eye movements. Journal ofVision, 10(3), 1–12.

Harrison, S. A., & Tong, F. (2009). Decoding reveals the contents of visual workingmemory in early visual areas. Nature, 458, 632–635.

Hayhoe, M. M., Shrivastavah, A., Mruczek, R., & Pelz, J. B. (2003). Visual memory andmotor planning in a natural task. Journal of Vision, 3(1), 49–63.

Helmholtz, H. von. (1867). Handbuch der Physiologischen Optik. Leipzig: LeopoldVoss.

Hochberg, J. (1968). In the mind’s eye. In R. Haber (Ed.), Contemporary theoryand research in visual perception (pp. 309–331). New York: Appleton CenturyCrofts.

Hollingworth, A., & Rasmussen, I. P. (2010). Binding objects to locations: Therelationship between object files and visual working memory. Journal ofExperimental Psychology: Human Perception and Performance, 36, 543–564.

von Holst, E., & Mittelstaedt, H. (1971). The principle of reafference: Interactionsbetween the central nervous system and the peripheral organs. In P.C. Dodwell(Ed. and Trans.), Perceptual processing: Stimulus equivalence and patternrecognition (pp. 41–71). New York: Appleton (Original work published 1950).

Horaguchi, T., & Sugino, K. (2008). Different memory types for generating saccadesat different stages of learning. Neuroscience Research, 55, 271–284.

Horowitz, T. S., & Wolfe, J. M. (1998). Visual search has no memory. Nature, 394,575–577.

Irwin, D. E. (1992). Memory for position and identity across eye movements. Journalof Experimental Psychology: Learning, Memory, and Cognition, 18(2), 307–317.

Irwin, D. E. (1996). Integrating information across saccadic eye movements. CurrentDirections in Psychological Science, 5(3), 94–100.

Irwin, D. E., McConkie, G. W., Carlson-Radvansky, L. A., & Currie, C. (1994). A localistevaluation solution for visual stability across saccades. Behavioral and BrainSciences, 17, 265–266.

Irwin, D. E., Yantis, S., & Jonidas, J. (1983). Evidence against visual integration acrosssaccadic eye movements. Perception & Psychophysics, 34, 35–46.

Irwin, D., Zacks, J. L., & Brown, J. S. (1990). Visual memory and the perception of astable visual environment. Perception & Psychophysics, 47(1), 35–46.

Jonides, J., Irwin, D. E., & Yantis, S. (1982). Integrating visual information fromsuccessive fixations. Science, 215, 192–194.

Kahneman, D., Treisman, A., & Gibbs, B. J. (1992). The reviewing of object files:Object specific integration of information. Cognitive Psychology, 24, 175–219.

Karn, K., Moeller, P., & Hayhoe, M. (1997). Reference frames in saccade targeting.Experimental Brain Research, 115, 267–282.

Klein, R. M. (2000). Inhibition of return. Trends in Cognitive Sciences, 4(4), 138–147.Klein, R. M., & MacInnes, W. J. (1999). Inhibition of return is a foraging facilitator in

visual search. Psychological Science, 10(4), 346–352.Kowler, E., Anderson, E., Dosher, B., & Blaser, E. (1995). The role of attention in the

programming of saccades. Vision Research, 35, 1897–1916.Lawrence, B. M., Myerson, J., & Abrams, R. A. (2004). Interference with spatial

working memory: An eye movement is more than a shift of attention.Psychonomic Bulletin & Review, 11(3), 488–494.

Liu, Y., Yttri, E. A., & Snyder, L. H. (2010). Intention and attention: Differentfunctional roles for LIPd and LIPv. Nature Neuroscience, 13(4), 495–502.

Macmillan, N. A., & Creelman, C. D. (2005). Detection theory: A user’s guide (2nd ed.).Mahwah, NJ: Erlbaum.

Matin, E. (1974). Saccadic suppression: A review and an analysis. PsychologicalBulletin, 81, 899–917.

McCarley, J. S., Wang, R. F., Kramer, A. F., Irwin, D. E., & Peterson, M. S. (2003). Howmuch memory does oculomotor search have? Psychological Science, 14(5),422–426.

McConkie, G. W., & Currie, C. B. (1996). Visual stability across saccades whileviewing complex pictures. Journal of Experimental Psychology: Human Perception& Performance, 22, 563–581.

McConkie, G. W., & Rayner, K. (1976). Identifying the span of the effective stimulusin reading: Literature review and theories of reading. In H. Singer & R. B. Ruddell(Eds.), Theoretical models and processes in reading (pp. 137–162). Newark, DE:International Reading Association.

McGuire, L. M. M., & Sabes, P. N. (2009). Sensory transformations and the use ofmultiple reference frames for reach planning. Nature Neuroscience, 12(8),1056–1063.

Moore, C. M., & Egeth, H. (1997). Perception without attention: Evidence ofgrouping under conditions of inattention. Journal of Experimental Psychology:Human Perception and Performance, 23, 339–352.

Müller, H. J., & von Mühlenen, A. (2000). Probing distractor inhibition in visualsearch: Inhibition of return. Journal of Experimental Psychology: HumanPerception and Performance, 26(5), 1591–1605.

332 I-Fan Lin, A. Gorea / Vision Research 51 (2011) 323–332

O’Regan, J. K. (1992). Solving the ‘‘real’’ mysteries of visual perception: The world asan outside memory. Canadian Journal of Psychology, 46, 461–488.

O’Regan, J. K., & Lévy-Schoen, A. (1983). Integrating visual information fromsuccessive fixations: Does trans-saccadic fusion exist? Vision Research, 23,765–768.

O’Regan, J. K., & Noe, A. (2001). A sensorimotor account of vision and visualconsciousness. Behavioral and Brain Science, 24, 939–973 [discussion 973–1031].

Palmer, J., Huk, A. C., & Shadlen, M. N. (2005). The effect of stimulus strength on thespeed and accuracy of a perceptual decision. Journal of Vision, 5(5), 376–404.

Pearson, D. G., & Sahraie, A. (2003). Oculomotor control and the maintenance ofspatially and temporally distributed events in visuo-spatial working memory.Quarterly Journal of Experimental Psychology, 56A(7), 1089–1111.

Pelli, D. G. (1997). The VideoToolbox software for visual psychophysics:Transforming numbers into movies. Spatial Vision, 10(4), 437–442.

Peterson, M. S., Kramer, A. F., Wang, R. F., Irwin, D. E., & McCarley, J. S. (2001). Visualsearch has memory. Psychological Science, 12(4), 287–292.

Pylyshyn, Z. (1989). The role of location indexes in spatial perception: A sketch ofthe FINST spatial-index model. Cognition, 32, 65–97.

Rayner, K., & Pollatsek, A. (1983). Is visual information integrated across saccades?Perception & Psychophysics, 34, 39–48.

Rensink, R. A. (2000). The dynamic representation of scenes. Visual Cognition, 7(1/2/3), 17–42.

Rojer, A. S., & Schwartz, E. L. (1990). Design considerations for a space-variant visualsensor with complex-logarithmic geometry. In Proceedings of the 10thinternational conference on pattern recognition (pp. 278–285). Washington, DC:IEEE Computer Society Press.

Schneider, W. X., & Deubel, H. (1995). Visual attention and saccadic eye movements:Evidence for obligatory and selective spatial coupling. In J. M. Findlay, R. W.Kentridge, & R. Walker (Eds.), Eye movement research: Mechanisms, processes andapplications (pp. 317–324). Amsterdam: Elsevier Science.

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

Sperry, R. W. (1950). Neural basis of the spontaneous optokinetic responseproduced by visual inversion. Journal of Computational Physiology Psychology,43, 482–489.

Stark, L., & Bridgeman, B. (1983). Role of corollary discharge in space constancy.Perception & Psychophysics, 34(4), 371–380.

Takeda, Y., & Yagi, A. (2000). Inhibitory tagging to continuous visual stimuli.Perception & Psychophysics, 62(5), 927–934.

Theeuwes, J., Belopolsky, A., & Olivers, C. N. L. (2009). Interactions between workingmemory, attention and eye movements. Acta Psychologica, 132(2), 106–114.

Todd, J. J., & Marois, R. (2004). Capacity limit of visual short-term memory in humanposterior parietal cortex. Nature, 428, 751–754.

Treisman, A., & Gormican, S. (1988). Feature analysis in early vision: Evidence fromsearch asymmetries. Psychological Review, 95, 15–48.

Treisman, A., & Sato, S. (1990). Conjunction search revisited. Journal of ExperimentalPsychology: Human Perception and Performance, 16, 459–478.

Umeno, M. M., & Goldberg, M. E. (2001). Spatial processing in the monkeyfrontal eye field: II Memory responses. Journal of Neurophysiology, 86,2344–2352.

Volkmann, F. C., Schick, A. M., & Riggs, L. A. (1968). Time course of visual inhibitionduring voluntary saccades. Journal of the Optical Society of America, 58(4),562–569.

Wheeler, M. E., & Treisman, A. M. (2002). Binding in short-term visual memory.Journal of Experimental Psychology: General, 131(1), 48–64.

Wolfe, J. M. (1998). What do 1000,000 trials tell us about visual search?Psychological Science, 9, 33–39.

Wolfe, J. M., Klempen, N., & Dahlen, K. (2000). Post-attentive vision. Journal ofExperimental Psychology: Human Perception and Performance, 26(2), 693–716.

Wong, E., & Mack, A. (1981). Saccadic programming and perceived location. ActaPsychologica, 48, 123–131.

Wurtz, R. H. (2008). Neuronal mechanisms of visual stability. Vision Research, 48,2070–2089.

Xu, Y., & Chun, M. M. (2006). Dissociable neural mechanisms supporting visualshort-term memory for objects. Nature, 440, 91–95.

Yeshurun, Y., & Schwartz, E. L. (1989). Shape description with a space-variantsensor: Algorithms for scan-path, fusion and convergence over multiple scans.IEEE Transactions on Pattern Analysis and Machine Intelligence, 11, 1217–1222.