**1. Introduction**

Knowledge about factors limiting visual perception is important to understand how we become aware of the world around us. This issue has often been studied by means of the rapid serial visual presentation task (RSVP) where target stimuli are embedded in background stimuli presented rapidly one after the other. A well-investigated phenomenon is the "attentional blink", which is the difficulty in discerning T2, the second of two target stimuli ("T1" and "T2") when presented at some critical moment after T1 [1,2]. The present study investigates another handicap to identifying T2: Some background

stimuli may resemble T2. Such "lure" stimuli, differing from the actual T2 only by occurring before T1 rather than afterwards, have been shown to impair T2 identification [3–9] and may be considered a model case for studying the time-dependent changes of relevance which the many objects in our environment may undergo, because the very same object that is irrelevant when encountered too early may become relevant later on. Several studies have provided evidence that the process launched by perception and associated rejection of lures is inhibition of detecting the targets [4–11].

In the studies by Niedeggen and colleagues [4–9] participants had to detect a color change of the fixation point to red (T1) and then immediately detect or discriminate the direction of coherent motion (T2) of dots surrounding fixation or, in [7], a flip of surrounding bars. Lags from T1 to T2 varied between 0 ms and 700 ms. Performance was nearly perfect for all measured lags (Figure 1 in [6]). Yet T2 identification drastically deteriorated at short T1–T2 lags (most at a lag of 0 ms) when lure stimuli were presented before T1. This negative effect was cumulative, increasing with the number of lures in a trial [5,9], well in line with the notion of a gradual increase of central inhibition [8].

**Figure 1.** Sequence of events in a trial. Participants had to identify the red letter (T1) and the blue digit (T2) embedded in a stream of background stimuli. At least nine pairs of background stimuli were presented before T1. In 80% of trials, three of these pairs contained stimuli that resembled T2 ("lures"), either by their blue color or by their being a digit. The three lures could be on the same side as the ensuing T2 (as in these examples) or on the other side.

In Zhang et al.'s studies [10,11], T1 was a red letter in a series of white letters at screen center, and T2 was a white digit, either also presented at center [10,11] or left or right [10]. Lures (i.e., white digits presented at center before T1) reduced T2 identification.

In Harris et al.'s study [12] a stream of object drawings was presented at screen center. Participants had to identify the two red objects (T1 and T2) among the other, black objects. The black object that was presented two frames before T1 could be the same object as the red T2. These lures had negative effects on T2 identification when T2 was presented briefly after T1. These negative effects turned to stable positive effects when lures and T2 were rotated or mirrored relative to each other.

In order to embed such time-dependent changes into a spatial dimension, in a previous study, we presented lure stimuli in dual-stream RSVP. In this task, first implemented by Holländer et al. [13] and used in our lab in several studies (from [14] onwards), two streams are presented, left and right from fixation, with T1 and T2 each occurring in either of these streams. In our standard version, background stimuli are black letters, T1 is a red letter, and T2 is a black digit. In our lure study in this set-up [15], lures, presented in half of the trials, were black digits like T2. This is similar to Zhang et al.'s [10,11] stimuli, except that they did not present two streams. In our paradigm with two streams, evidence was found for both positive and negative priming exerted by lures on T2 identification, formally similar to Harris et al.'s [12] results. Specifically, in our task [15] lures had negative e ffects on T2 identification when the lag between T1 and T2 was 3 frames, This e ffect was shifted towards positive priming when lures and T2 were in the same stream and when the intervening T1 was "out of the way" in the other stream and when one of three lures was identical to T2. This latter result apparently di ffers from the above-reported results by Niedeggen et al. and Harris et al. [4–9,12] where identity of lures and T2 impeded rather than facilitated T2 identification.

Because of these conflicting results, here we reasoned that both mechanisms might be e ffective, either one being triggered by specific relations between lures and T2. To detail, similarity of lures to T2 might result in positive priming of lure contents for discerning T2 [12,15] whereas the property of lures to capture attention due to their similarity to T2 might inhibit and impede T2 identification. In the present study, we aimed at separating these two presumed e ffects by assigning two di fferent features to T2 and using two types of lures: Unlike in previous studies [10,11,15], the T2 targets stood out not only by being digits but also by their color, being blue rather than black. In di fferent trials, lures were either three same-category lures, i.e., black digits, or three attention-catching, salient lures, i.e., blue letters. We assumed that digit lures will positively prime T2 identification, whereas salient lures, being only superficially similar to T2, will not be able to prime T2 identification but will rather distract attention and, thereby, impede T2 identification. To optimize the presumed distinction of effects, one of the three digit lures in a trial was identical to T2, like in half the lure trials of our previous study [15]. Positive priming from lures on T2 might be considered trivial when one of the lures is T2. But negative identity priming from lures on T2 has been shown as well [12], and moreover, negative priming from identical irrelevant to relevant stimuli when separated by masks or other stimuli has often been demonstrated [16–18].

Sample trials of the three conditions no-lures, digit lures, color lures are depicted in Figure 1.

#### *1.1. Stream Change from Lures to T2*

The supposed attention-distracting e ffect of salient color lures may involve a strong spatial component. I.e., the impeding e ffect of salient lures on T2 identification may become evident only if lures and T2 are in di fferent streams, because lures will have attracted attention to their stream which then turns out to be the incorrect one. Correspondingly, no negative e ffect, or even some positive e ffect on T2 identification, is expected if lures and T2 are in the same stream because, then, the lures will have attracted attention to the correct stream.

In contrast, by being content-dependent, the supposed positive priming e ffect of digit lures on T2 identification might be less tightly linked to spatial conditions. Thus, this positive e ffect should not di ffer much between stream change and stream continuity from lures to T2.

#### *1.2. Role of T1 Stream*

The preceding considerations were made without taking the relevant and attention-catching stimulus between lures and T2 into account, which is T1, the red letter. In [15], some evidence was obtained for the assumption that positive priming e ffects built up by lure presentations are broken and even inverted by T1 presentation (signaling that the preceding digits were too early to be T2). To keep things simple, it may at least be expected here that the spatial position of T1 will have e ffects on spatially specific e ffects: The attention-distracting e ffects of salient lures on other-stream T2 might be enhanced by T1 occurring in the lure stream and attenuated by T1 occurring in the T2 stream.

## *1.3. Event-Related Potentials*

Event-related potentials (ERPs) were recorded in order to provide additional evidence on di fferential processing of same-category digit lures and salient color lures. This additional evidence was expected to be particularly relevant to evaluate immediate processing of the lures because there

were only indirect behavioral measures of lure processing (by their moderating e ffects on discerning T2 and T1).

## *1.4. Lure-Evoked ERPs*

When a trial included lures, there were always three of them, occurring in the same stream. If lures evoke inhibition then this inhibition should accumulate across the three lures. Lure-evoked ERPs had been measured by Niedeggen et al. [8,9,17], Verleger et al. [13], and Zhang et al. [6]. In all studies, a negative ERP component was evoked by lures at anterior scalp sites around 300 ms. In Niedeggen et al.'s studies [7,8,19], this component became larger across the three lure positions within a trial. (Zhang et al. [10] presented only one lure per trial at a fixed position, immediately before T1). Not occurring in control conditions (where lures did not share features with the target), the increase was interpreted by Niedeggen et al. [7,8,19] as reflecting frontal "gating" against being prematurely activated by these stimuli. However, this increase was not replicated in Verleger et al.'s [15] study. To accommodate this divergence, we assume for the present study that such frontal negativity will not build up with the same-category digit lures but will do so with the distracting salient color lures.

In addition, our use of bilateral streams allowed for measuring the N2pc component. N2pc is a negative deflection around 250 ms after onset of laterally presented relevant stimuli, at contralateral sites above the visual cortex [20,21] reflecting spatially selective processing of these stimuli [22–24]. Measuring N2pc evoked by lures provides a test for the assumption that color lures will attract more attention than digit lures: If this is true, N2pc amplitudes are expected to be larger with color lures than with digit lures, particularly with the first of the three lures because when the first color lure has attracted attention to its side the focus of attention may stay on that side such that the further lures do not require a shift of attention any more.

#### *1.5. T1-Evoked ERPs*

Being a relevant and salient lateral event, the red-colored T1 evokes N2pc in this dual-stream task [12,23]. One might speculate that the preceding sequence of three lures is helpful to expecting T1 precisely in time, so T1-evoked N2pc might increase in lure trials. However, no such e ffect was obtained in our previous study [13]. What is to be expected, though, is that the shifting of attention by preceding color lures will a ffect T1-evoked N2pc. Without lures, attention will be distributed across the two streams. Color lures will attract attention to their stream. T1 will require less shifting of attention when presented in the same stream as color lures and will require more shifting of attention when color lures had occurred in the other stream. Accordingly, T1-evoked N2pc is expected to be smaller than in trials without lures when color lures and T1 are in the same stream, and to be larger when color lures and T1 are in di fferent streams.

## *1.6. T2-Evoked ERPs*

Likewise, being a relevant event, T2 will evoke N2pc, too [10,14,25,26]. In Zhang et al.'s study [10], N2pc was delayed when lures preceded. In contrast, in Verleger et al.'s study [15], N2pc amplitudes were reduced, but only when lures and T2 were in the same stream. It may be speculated that the possibly inhibition-related e ffect of [10] might become evident in the present study after color lures while the possibly priming-related e ffect in [15] might become evident after digit lures.

To summarize, the major purpose of the present study was to obtain divergent e ffects of the two types of lures on T2 identification, and the ERPs evoked by lures, T1, and T2 were expected to be helpful in describing the mechanisms of these lure e ffects.

#### **2. Materials and Methods**
