Masked stimuli modulate endogenous shifts of spatial attention

Abstract

Unconscious stimuli can influence participants’ motor behavior but also more complex mental processes. Recent research has gradually extended the limits of effects of unconscious stimuli. One field of research where such limits have been proposed is spatial cueing, where exogenous automatic shifts of attention have been distinguished from endogenous controlled processes which govern voluntary shifts of attention. Previous evidence suggests unconscious effects on mechanisms of exogenous shifts of attention. Here, we applied a cue-priming paradigm to a spatial cueing task with arbitrary cues by centrally presenting a masked symmetrical prime before every cue stimulus. We found priming effects on response times in target discrimination tasks with the typical dynamic of cue-priming effects (Experiments 1 and 2) indicating that central symmetrical stimuli which have been associated with endogenous orienting can modulate shifts of spatial attention even when they are masked. Prime–Cue Congruency effects of perceptual dissimilar prime and cue stimuli (Experiment 3) suggest that these effects cannot be entirely reduced to perceptual repetition priming of cue processing. In addition, priming effects did not differ between participants with good and poor prime recognition performance consistent with the view that unconscious stimulus features have access to processes of endogenous shifts of attention.

Introduction

Everyday people are faced with an incoming stream of information about their environment that is too vast to be processed in every detail. Attention can be defined as the mechanism by which people select behaviorally relevant information out of this stream for further processing. Information can be selected according to different features of the input. One mode of attention that has been studied extensively is the deployment of attention across visual space. Frequently, spatial attention is studied in the spatial cueing paradigm (e.g. Posner, 1980). In a typical spatial cueing experiment participants have to detect or identify visual target stimuli that can appear at different locations on a screen. On each trial, a cue stimulus informs participants about the likely location of the subsequent target stimulus (e.g. on the left or right side of the screen). When the target is presented at the predicted location, i.e. the cue is valid, participants show better detection or identification performance than when the cue is invalid and the target is presented at a different location. Cue stimuli can be presented in a variety of ways. Corresponding to the effects of peripherally and centrally presented cues, an early distinction has been made between exogenously and endogenously controlled orienting of attention, respectively (Jonides, 1981, Posner, 1980).

Effects of peripheral cues have been called “exogenous” because evidence suggested that these cues automatically attract attention rather independently of participants’ intentions. Evidence for the involuntary, reflexive nature of these shifts of attention arose from the finding that these effects occur even when cues are non-predictive or even counter-predictive for target locations (Posner, Cohen, & Rafal, 1982). In contrast, effects of centrally presented cues have traditionally been termed “endogenous” because it was assumed that these effects are based on controlled processing which takes current goals and intentions into consideration. The “voluntary” nature of these shifts of attention has been derived from the idea that these shifts occur only when cues are informative as to the location of the following target stimulus and participants make active use of the cues because participants are informed about the cues’ predictiveness. In the following the term “voluntary” shifts of attention is synonymous to shifts of attention that correspond to the information of predictive cues, whereas “involuntary” shifts of attention refer to shifts that result when cues are non-predictive.

It has been proposed that exogenous and endogenous shifts of attention result from different mechanisms, although both mechanisms can operate at the same time (Jonides, 1981, Müller and Rabbitt, 1989, Prinzmetal et al., 2005). Only endogenous orienting is thought to require intention and cognitive control, because the cue stimulus and the target are shown at different locations and thus, participants have to move their attention intentionally away from the cue to the expected target location (McCormick, 1997). Moreover, intention and cognitive control both have been associated with consciousness (e.g. Dehaene and Naccache, 2001, Jack and Shallice, 2001).

Several recent empirical findings, however, have challenged the traditional distinction between non-predictive peripheral cues that induce reflexive, involuntary shifts of attention due to automatic, exogenously triggered processes on the one hand, and predictive centrally presented cues that induce intentional, voluntary shifts of attention due to controlled, endogenously governed processes on the other hand (e.g., Kingstone, Smilek, Ristic, Friesen, & Eastwood, 2003). In an fMRI study Peelen, Heslenfeld, and Theeuwes (2004) found that shifts of attention to central and peripheral cues recruit similar neural mechanisms. The distinction is also questioned by validity effects of peripheral cues on shifts of spatial attention. For instance, Posner et al. (1982) used peripheral cues that were predictive for a target at another location and found that participants were able to use the information conveyed about the likely target location to shift their attention voluntarily away from the cue. On the other hand special types of centrally presented cues that lack any predictive value can also trigger shifts of attention rather automatically. For instance, spatially compatible cues such as centrally presented arrows or eye gaze stimuli (e.g. Friesen and Kingstone, 1998, Tipples, 2002) can elicit shifts of spatial attention even when the cues are not predictive for target locations (Ristic & Kingstone, 2006). These effects suggest that shifts of spatial attention in response to these centrally presented cues are also susceptible to overlearned, involuntary, automatic processes. The assumption that orienting to centrally presented arrow cues is an endogenous voluntary process is further questioned by a recent study by Risko and Stolz (2010) in which implicit learning seemed to play an important role in the orienting of attention to the cues, rather than participants’ insights into the validity of cues in the current context.

Lambert, Roser, Wells, and Heffer (2006) have recently suggested that a distinction should be made regarding whether central or peripheral stimuli that serve as cues enable spatial correspondence learning rather than assuming different mechanisms to account for spatial cueing effects with central and peripheral cues. These authors assume that it is important whether stimuli enable an associative learning of spatial correspondences between the cue display and the target display. Spatial correspondence can be learned rather easily with asymmetrical central and peripheral cues. A weak form of spatial correspondence learning can also occur with bilateral symmetrical stimuli like two Ts or two Xs that are concurrently presented left and right of fixation (e.g., Lambert and Duddy, 2002, Shin et al., 2011). In contrast, however, when a single symmetrical stimulus like T or X serves as a predictive cue, spatial correspondence learning should not occur. This view has been supported by the absence of spatial cueing effects with these stimuli (Lambert et al., 2006, Shin et al., 2011). In sum, these challenging findings call for an elaboration of the characteristics of the processes that are engaged in endogenous and exogenous shifts of attention. This should also incorporate the role of cue stimuli that enable spatial correspondence learning and symmetrical stimuli that do not, which might correspond to exogenously and endogenously governed shifts of attention, respectively. In the present study we used centrally presented symmetrical cue stimuli to examine endogenous shifts of spatial attention and the issue whether such shifts of attention depend on visual awareness of the effective features of cue stimuli.

Attention and consciousness are somewhat related, yet distinct concepts (Koch & Tsuchiya, 2007). Several studies have attempted to investigate effects of unconscious cues on shifts of spatial attention. The visibility of cue stimuli was reduced by different methods in different studies, either by reducing size or luminance contrast of the cues, or by the subsequently presentation of masking stimuli. A series of studies suggests that spatially compatible cues can affect shifts of spatial attention even when the effective cues are masked in a way that participants are not aware of them. This has been reported for peripheral (Kentridge et al., 1999, Lambert et al., 1999, McCormick, 1997, Mulckhuyse et al., 2007, Scharlau and Ansorge, 2003) and central cues (Cole and Kuhn, 2010, Reuss et al., 2011b, Sato et al., 2007). For a review see Mulckhuyse and Theeuwes (2010).

Of particular relevance is the study by McCormick (1997), who contrasted the effects of predictive and non-predictive peripheral cues. Cues predicted a target at the opposite side of the screen. Luminance contrast of the cues was varied so that some cues were above and some below a subjective threshold of awareness. Although participants reported that they had not seen the cues, McCormick (1997) found involuntary shifts of attention to the location of the low-contrast cue. In contrast, when participants reported to see the cue they were able to use the information conveyed about the likely target location to shift their attention voluntarily away from the cue. The author concluded from this that endogenous orienting requires executive control which depends on the awareness of cue stimuli.

All of the studies on effects of invisible cues on spatial attention mentioned above used spatially compatible stimuli that can be linked to mechanisms which govern involuntary shifts of attention. To our knowledge, there are currently no reports of corresponding effects of centrally presented symmetrical cues. Therefore, we considered it worthwhile to investigate whether comparable effects with masked centrally presented symmetrical cues could be found. Central symmetrical cues are special in that they do not enable spatial correspondence learning (Shin et al., 2011) and they differ from arrows and eye gaze stimuli because they have no overlearned associations to a spatial direction. Therefore, centrally presented symmetrical cues require an interpretation of their meaning which has been associated to endogenous control of spatial attention. We reasoned that an effect of unconscious stimulus features of central symmetrical stimuli on shifts of spatial attention could be taken as direct evidence for the view that the mechanisms which govern endogenous shifts of visual attention are susceptible to the effects of unconscious information.

The studies mentioned above potentially challenge the traditional distinctions between involuntary, exogenous shifts of attention that are triggered by peripheral cues and voluntary, endogenous shifts caused by central cues and also challenge the corresponding links to automatic and controlled mechanisms that are influenced by unconscious and conscious stimuli, respectively. Presently, however, there is no convincing evidence for effects of unconscious stimuli on mechanisms of endogenous shifts of attention. To the extent that endogenous shifts of attention are governed by executive control processes, it could be assumed that these control processes require consciously accessible stimuli. Recently, however, this notion has been called into question both theoretically (Hommel, 2007) and empirically in studies that showed effects of unconscious stimuli on processes of task preparation (Lau and Passingham, 2007, Mattler, 2003b, Mattler, 2005, Mattler, 2007) and response inhibition (Krüger et al., 2013, van Gaal et al., 2008, van Gaal et al., 2009).

For instance, Mattler, 2003b, Mattler, 2005 showed priming effects of metacontrast-masked stimuli in several cueing tasks thought to involve executive control processes. In Experiment 3 of his study participants were instructed to shift their attention to either the auditory or visual modality according to a symbolic cue stimulus (a square or a diamond). Unbeknownst to them, this cue stimulus was always preceded by a masked prime which was also either a square or a diamond. When the cue and the prime had the same shape, participants were faster to identify a target stimulus in the cued modality than when they were of different shapes. This was interpreted as evidence that unconscious stimulus features can affect shifts of attention between modalities. On the one hand, spatial attention has been distinguished from other kinds of attention (Hillyard & Anllo-Vento, 1998) consistent with the view that space is a “special” stimulus dimension for the control of visual attention (e.g., Treisman, 1998, Treisman and Gelade, 1980). On the other hand, recent evidence suggests that orienting of spatial attention shares some mechanisms with other kinds of attentional preparation similar to those examined in previous cue-priming studies (Chiu and Yantis, 2009, Slagter et al., 2005). Therefore, we thought it interesting to transfer Mattler’s cue-priming paradigm to a spatial cueing task. Given that previous cue-priming studies mostly used central symmetrical cues, such a transfer might help to clarify the issue whether unconscious features of centrally presented symmetrical stimuli can affect spatial attention.

The present study was designed to employ a cue-priming paradigm in the domain of spatial attention. To the extent that performance in a spatial attention task is affected by unconscious cue-priming in the same way as in other tasks, priming effects should exhibit similar time courses when stimulus onset asynchronies (SOAs) between the critical stimuli are varied. This could increase the amount of evidence for the view that cue-priming is a domain independent phenomenon. Beyond this, we aimed to employ stricter objective measures of prime visibility than previous studies that investigated effects of masked central cues (Cole and Kuhn, 2010, Sato et al., 2007). In addition, we thought it is important to control for eye movements to rule out the possibility that effects result from priming of oculomotor processes. The premotor theory of attention (Rizzolatti, Riggio, Dascola, & Umilta, 1987) proposes that spatial attention overlaps with the programming of saccades. If spatial attention is indeed based on subthreshold saccade activation, cue-priming effects in a spatial cueing task could possibly be explained solely on a motor level. In this case priming of spatial attention could be more similar to motor than to non-motor priming (Mattler, 2003b). To control for such motor effects we tracked eye movements in all experiments and excluded all trials on which eye movements occurred.