Consciousness and Cognition

Transcription

1 Consciousness and Cognition xxx (2009) xxx xxx Contents lists available at ScienceDirect Consciousness and Cognition journal homepage: Review Prior-entry: A review Charles Spence a, *, Cesare Parise a,b a Crossmodal Research Laboratory, Department of Experimental Psychology, University of Oxford, United Kingdom b Department of Cognitive and Education Science, University of Trento, Italy article info abstract Article history: Received 13 September 2009 Available online xxxx Keywords: Prior entry Attention TOJ SJ Crossmodal Spatial The law of prior entry was one of E.B. Titchener s seven fundamental laws of attention. According to Titchener (1908, p. 251): the object of attention comes to consciousness more quickly than the objects which we are not attending to. Although researchers have been studying prior entry for more than a century now, progress in understanding the effect has been hindered by the many methodological confounds present in early research. As a consequence, it is unclear whether the behavioral effects reported in the majority of published studies in this area should be attributed to attention, decisional response biases, and/or, in the case of exogenous spatial cuing studies of the prior-entry effect, to sensory facilitation effects instead. In this article, the literature on the prior-entry effect is reviewed, the confounds present in previous research highlighted, current consensus summarized, and some of the key questions for future research outlined. In particular, recent research has now provided compelling psychophysical and electrophysiological evidence to support the claim that attending to a sensory modality, spatial location, or stimulus feature/attribute can all give rise to a relative speeding-up of the time of arrival of attended, as compared to relatively less attended (or unattended) stimuli. Prior-entry effects have now been demonstrated following both the endogenous and exogenous orienting of attention, though prior-entry effects tend to be smaller in magnitude when assessed by means of participants performance on SJ tasks than when assessed by means of their performance on TOJ tasks. Ó 2009 Published by Elsevier Inc. 1. Introduction Does attention speed-up perceptual processing? Or, in other words, does attending to (or expecting) a particular stimulus (or event) mean that it will be perceived earlier in time than if attention had been directed elsewhere? This seemingly simple question is in fact one of the oldest in the field of experimental psychology (Mollon & Perkins, 1996; Scharlau, 2007; see Spence, Shore, & Klein, 2001, for a review). However, while researchers have been investigating the topic of temporal perception in humans for more than two centuries, it is only in the last decade or so that convincing psychophysical evidence in support of the prior-entry effect (as the phenomenon is known) has finally been obtained (see Shore & Spence, 2005). That said, there has been a recent resurgence of research interest in the prior-entry effect (e.g., Lester, Hecht, & Vecera, 2009; Weiss & Scharlau, 2009; West, Anderson, & Pratt, in press; Yates & Nicholls, 2009; Zhuang & Papathomas, 2009). What is more, the latest research utilizing event-related potentials (ERPs) has now started to demonstrate just how early in human information processing the effects of attention can be observed (McDonald, Teder-Sälejärvi, Di Russo, & Hillyard, 2005; Vibell, Klinge, Zampini, Spence, & Nobre, 2007, submitted for publication). * Corresponding author. Address: Department of Experimental Psychology, University of Oxford, South Parks Road, Oxford, OX1 3UD, United Kingdom. Fax: address: charles.spence@psy.ox.ac.uk (C. Spence) /$ - see front matter Ó 2009 Published by Elsevier Inc. doi: /

2 2 C. Spence, C. Parise / Consciousness and Cognition xxx (2009) xxx xxx The findings of research on the prior-entry effect are not only of interest to psychologists, psychophysicists, and cognitive neuroscientists, but are also of relevance to philosophers interested in the question of how time (at least the fine millisecond timescale captured by studies of prior entry; see Buonomano & Karmarkar, 2002; Eagleman, 2008; Eagleman et al., 2005) is represented neurally (e.g., see Dennett & Kinsbourne, 1992; Durgin & Sternberg, 2002; Kelly, 2005; Mellor, 1985; Roache, 1999). Indeed, the latest research on the prior-entry effect has shown that information concerning temporal order is, to some extent, represented temporally in the brain (at least for the case of crossmodal temporal judgments; Köhler, 1947; Vibell et al., submitted for publication, 2007). In this article, we review the extensive empirical literature that has investigated the effects of attention (to a sensory modality or to a spatial location) on temporal perception (in both unimodal and multisensory settings) in humans. 2. Measuring the effect of attention on temporal perception The problem when investigating the effects of attention on temporal perception is that it is impossible for a person to index when exactly a given stimulus or event was perceived as occurring. Instead, researchers have had to rely on a person s judgments of the relative timing of an event of interest with respect to another (comparison or marker) stimulus (see also Schneider & Bavelier, 2003). The two tasks that have been used most frequently to study the effects of attention on temporal perception in humans are the temporal order judgment (TOJ) task and the simultaneity judgment (SJ) task. For both tasks, the stimulus onset asynchrony (SOA) between the two to-be-judged stimuli on each trial is normally varied using the method of constant stimuli (e.g., Spence, Shore, & Klein, 2001), or else some form of adaptive procedure (e.g., Stelmach & Herdman, 1991; Sternberg, Knoll, & Gates, 1971; Zampini, Guest, Shore, & Spence, 2005). In a typical TOJ study, participants have to report which stimulus was presented first (or, on occasion, second; see Frey, 1990; Parise & Spence, 2008, 2009; Shore, Spence, & Klein, 2001), while in the SJ task, they have to report whether the two stimuli were presented simultaneously or not. Occasionally, these two tasks are combined into a so-called ternary-response task (Jaśkowski, 1993; Stelmach & Herdman, 1991; Stone, 1926; Ulrich, 1987; Zampini et al., 2007), in which the participants either report which stimulus was presented first, or else that the two stimuli appeared to have been presented simultaneously. While the ternary-response task might, at first, seem preferable in terms of providing a more nuanced measure of people s perceptual experience, it should be noted that the task suffers from the problem that participants vary markedly in the criterion they set to choose the simultaneous response option when given three (rather than just two) response alternatives (e.g., see Schneider & Bavelier, 2003). This can make it rather difficult to compare the results of different studies (e.g., see Jaśkowski, 1993; and Stelmach & Herdman, 1991, on this point). Hence, the majority of prior entry studies published to date have tended to use TOJ or SJ tasks instead, although it should be noted that all three methods are subject to criterion effects of one sort or another. Two key performance indicators can be estimated by fitting psychometric (e.g., cumulative Gaussian) functions to the data from TOJ tasks: the first is the amount of time by which one stimulus has to precede (or follow) the other in order for the two stimuli to be perceived as simultaneous, known as the point of subjective simultaneity (PSS; note that this value actually reflects an estimate of the SOA at which participants would be likely to make each response equally often). The second is a participant s sensitivity to the temporal order in which two stimuli were presented, known as the just noticeable difference (JND). The JND is defined as the delay between two stimuli needed for participants to perceive the correct order of presentation of the two stimuli at a specified level. In studies of the prior-entry effect, the JND has conventionally been calculated as half the temporal interval between the 25% and 75% points on the psychometric function. Similarly, when analyzing the data from a SJ task, researchers typically fit a simple Gaussian function to the simultaneous response data. The spread (or standard deviation, SD) of this distribution provides a measure of a participant s sensitivity to asynchrony (equivalent to the JND) while the center of the distribution provides an estimate of the PSS. The data from the ternary-response task has been analyzed in different ways in different studies, but often involves some combination of the above techniques (e.g., Zampini et al., 2007). No matter what the task (i.e., TOJ, SJ, or ternary-response task), evidence supporting the existence of the prior-entry effect comes from the observation of a significant difference in the PSS between those conditions in which one of the target stimuli is attended as compared to when the other stimulus is attended instead. According to Titchener s (1908) claim regarding prior entry, the stimulus that is attended should be presented later in time relative to the unattended stimulus in order for the two stimuli to be perceived as simultaneous (i.e., at the PSS). It is, however, important to note that the TOJ, SJ, and ternary-response tasks are all subject to the potential influence of a variety of response biases. Crucially, certain kinds of response bias that can affect performance in the TOJ task may be expected to lead to a shift in the PSS that can easily be confused with an attentional (i.e., perceptual) effect on the PSS. So, for example, participants may simply report the modality to which they had been instructed to attend (rather than the stimulus that was perceived as having been presented first). The shift in the PSS documented in such studies is, however, decisional, rather than attentional, in nature. To date, no evidence has been forthcoming regarding potential biases that might influence the measure of the PSS derived from fitting a psychometric function to the data from the SJ task. Consequently, many researchers now argue that it is more appropriate to use the SJ task than to use TOJs when trying to assess the effects of attention on the perceived time of arrival of an event (i.e., when trying to measure the effects of prior entry on temporal perception; see Schneider and Bavelier (2003), Zampini, Shore, and Spence (2005); though see below).

3 C. Spence, C. Parise / Consciousness and Cognition xxx (2009) xxx xxx 3 It is worth noting here that researchers have yet to come to a consensus with regard to the question of whether SJs and TOJs actually measure the same thing or not: the fact that various experimental manipulations have been shown to have different effects on SJs and TOJs (see Guerrini, Berlucchi, Bricolo, & Aglioti, 2003; Shore, Gray, Spry, & Spence, 2005; Vatakis, Navarra, Soto-Faraco, & Spence, 2008), has led some researchers to argue that they may measure somewhat different aspects of temporal perception, and even that they perhaps reflect the operation of different underlying neural mechanisms (e.g., Allan, 1975; Jaśkowski, 1991; Mitrani, Shekerdjiiski, & Yakimoff, 1986; Stelmach & Herdman, 1991; Van Eijk, Kohlrausch, Juola, & van de Par, 2008). Theoretically-speaking, attention might be expected to have at least two distinct effects on temporal perception: first, attending to a stimulus (i.e., to its location, modality, etc.) might lead to the prior entry (or earlier arrival) of the attended stimulus relative to the same stimulus when a participant s attention happens to have been directed elsewhere. 1 Additionally, however, attention might also be expected to improve the precision of participants judgments for attended stimuli in both the TOJ and SJ task (analogous to the enhancement of spatial resolution documented in studies of visual spatial attention; e.g., Carver & Brown, 1997; Nicol, Watter, Gray, & Shore, 2009; Pestilli, Viera, & Carrasco, 2007; Rolke, Dinkelbach, Hein, & Ulrich, 2008; Shore et al., 2001; Stelmach & Herdman, 1991; Yeshurun & Carrasco, 1998; Yeshurun & Levy, 2003). The latest evidence has shown that both spatial and temporal attentional manipulations can lead to improved temporal-discrimination performance (e.g., Chica & Christie, 2009; Correa, Sanabria, Spence, Tudela, & Lupiáñez, 2006), though it should be noted that the story here is complicated by the fact that attention can both increase and decrease precision (see Nicol et al., 2009; Yeshurun & Levy, 2003). However, the shift in the PSS (rather than any reduction in the JND) is the performance measure that is most directly relevant to assessing Titchener s (1908) claims regarding the existence of the prior-entry effect, and it is on that we will focus. It is also worth noting that while the participants in studies of the prior-entry effect are normally encouraged to make unspeeded perceptual judgments (though see Gibson and Egeth (1994), Prinzmetal, McCool and Park (2005), Prinzmetal, Park and Garrett (2005), Santangelo and Spence (2009), for exceptions), analysis of the distribution of reaction times (RTs) from such studies can also prove to be informative (Cardoso-Leite, Gorea, & Mamassian, 2007; Heath, 1984): so, for example, if a wide enough range of SOAs are tested, one may find that participants respond more slowly at SOAs close to the PSS and more rapidly at SOAs that are further from the PSS (see Shore et al., 2001). Results such as these are consistent with participants responding most slowly when they are least certain with regard to the appropriate response, although, of course, it should be noted that the causes of differences in RT are often difficult to ascertain unequivocally (e.g., Blake, Land, & Mollon, 2008; McDonald, Teder-Sälejärvi, & Hillyard, 2000; Watt, 1991). Shore et al. (2001) highlighted the existence of just such a shift of the peak of the RT distribution in their study of visual spatial prior entry. However, while there are a number of ways of measuring the effects of attention on temporal perception, the focus in this review (in line with the majority of studies that have been published on the prior-entry effect) will be on any shift in the PSS derived from participants temporal judgments. 3. Prior-entry effects resulting from attention being directed to a sensory modality: experimental evidence Psychologists distinguish between the endogenous and exogenous orienting of spatial attention (e.g., Corbetta & Shulman, 2002; Klein, 2004; Klein & Shore, 2000; Prinzmetal, McCool et al., 2005; Prinzmetal, Zvinyatskovskiy, Gutierrez, & Dilem, 2009). Exogenous shifts of attention can be elicited by the peripheral presentation of a non-predictive cue stimulus, whereas endogenous shifts of attention are voluntarily induced by the provision of prior information about the likely identity or location of the target. Several early studies provided evidence that is consistent with the claim that endogenously (i.e., voluntarily) attending to a particular sensory modality results in the prior entry into awareness of stimuli subsequently presented in that modality (Frey, 1990; Sternberg et al., 1971; Stone, 1926; Wilberg & Frey, 1977, 1990). In these studies, a pair of stimuli was presented on each trial and an attempt made to direct the participant s attention to the modality of one or other target stimulus (by default, the other stimulus in such studies is considered as being relatively unattended ). For example, Stone observed a 46 ms prior-entry effect (see Sternberg & Knoll, 1973) when participants attention was directed to either the auditory or tactile modality; on each trial, the participants judged whether a tactile or an auditory stimulus had been presented first (see Table 1). Sternberg et al. in an oft-cited (though never published) conference presentation given at a meeting of the Psychonomics Society in 1971, reported a 55 ms prior-entry effect when participants attention was directed to either audition or touch (once again, only stimuli in these two modalities were presented to participants). In a second experiment, Sternberg and his colleagues documented a 30 ms prior-entry effect when participants attention was directed to the auditory or visual modalities instead (note that in this latter study, the participants had to judge on each trial whether an auditory or visual stimulus had been presented first). Importantly, however, many other studies have failed to show any such prior-entry effect (see Cairney, 1975a; Drew, 1896; Frey & Wilberg, 1975; Hamlin, 1895; Vanderhaeghen & Bertelson, 1974). What is more, these null results would appear to reflect more than simply a lack of statistical power (cf. Frick, 1995), since the numbers of participants/trials etc. tested were fairly similar across the various studies that either have, or have not, demonstrated a prior-entry effect. In fact, the inconsistent pattern of results reported in the early research led Pashler (1998, p. 260) to conclude that the empirical 1 While it is often claimed that attention speeds up the perceptual processing of attended (or expected) stimuli, it is important to note that in addition or even instead of this facilitation, the perceptual processing of unattended stimuli might also be delayed (see Spence, Shore, & Klein, 2001, Fig. 14, on this point).

4 4 C. Spence, C. Parise / Consciousness and Cognition xxx (2009) xxx xxx Table 1 Summary of published studies that have shown a significant prior-entry effect resulting from the endogenous focusing of a participants attention on a particular sensory modality. The studies have been separated into four groups as a function of the participants task: non-orthogonal TOJ, Orthogonal TOJ, SJ, or ternary-response task (A = Auditory; T = Tactile; V = Visual; P = Pain; the modality here referring both to the stimulus modalities presented and the sensory modalities attended). Note that in Zampini et al. s (2007) study the prior-entry effect was calculated on the basis of the average of the prior-entry effect reported in the SJ data (40 ms) and that reported by looking at the TOJ responses (22 ms). Task Study Magnitude of prior-entry effect (in ms) Target modalities TOJ (non-orthogonal) TOJ (orthogonal) SJ Ternary-response task Stone (1926) 46 AT Sternberg et al. (1971) 55 AT 30 AV Spence et al. (2001; E2) 121 VT Vibell et al. (2007) 38 VT Zampini et al. (2005b) 17 AV Zampini et al. (2007) 29 VP evidence in support of the phenomenon of prior entry was unconvincing. Spence, Shore, & Klein, 2001 were, however, able to highlight a number of methodological and interpretational problems with all previous studies of the prior-entry effect both the significant prior-entry effects reported by certain researchers and the null effects reported by others. For instance, Spence, Shore, and Klein (2001); see also Cairney (1975a) pointed out that the apparent prior-entry effects reported in many early studies might simply have reflected some form of response bias, given that participants attention was manipulated along the same dimension that they had to use when responding (i.e., participants were instructed to attend to modality X or to modality Y, and they had to judge whether the stimulus in modality X or Y had been presented first): that is, given the sometimes-difficult perceptual task of judging which of two near-simultaneous events had actually been presented first, participants might simply have chosen to respond on the basis of the stimulus that they had been instructed to attend to when uncertain of the correct response. Clearly, this form of decisional bias needs to be distinguished from the perceptual change that Titchener (1908) had in mind when he first outlined the concept of prior entry (see also Jaśkowski, 1993). 2 On the other hand, Spence et al. highlighted the possibility that the null results reported in many other studies of the prior-entry effect (Cairney, 1975a; Drew, 1896; Frey & Wilberg, 1975; Hamlin, 1895; Vanderhaeghen & Bertelson, 1974) might have reflected the failure by the experimenters concerned to manipulate their participants attention successfully (see also below) Orthogonal-response paradigm Spence, Shore, and Klein (2001) introduced the orthogonal-responding version of the TOJ task in order to reduce the likelihood that decisional response bias would affect participants TOJ performance. Specifically, in their studies, attention was manipulated in one dimension (i.e., the participants were instructed to attend vision or to attend touch ) while the participants had to judge the temporal order of the stimuli along another (orthogonal) dimension (e.g., Was the first stimulus presented on the left versus on the right? ). The visual and tactile stimuli in Spence et al. s studies were always presented from the same set of spatial locations (with one possible target location situated on either side of fixation). One target was always presented from the left and the other stimulus from the right. In order to ensure that their participants attention was indeed endogenously focused on either the visual or tactile modality, Spence et al. manipulated the probability of target stimuli being presented in each modality. In particular, they varied the relative proportions of unimodal tactile and visual TOJ trials in the different blocks of experimental trials. So, for example, the majority of stimuli (75%) in the attend touch blocks were tactile (as compared to just 25% visual stimuli), while these stimulus probabilities were reversed in the attend vision blocks (see Fig. 1). What is more, the participants were informed verbally of the stimulus probabilities at the start of each block of experimental trials, and they were explicitly instructed to direct their attention to the more probable target modality. Spence, Shore, and Klein (2001) results demonstrated a significant prior-entry effect: that is, the visual target had to be presented further in advance of the tactile stimulus in their study in order for the two stimuli to be perceived as simulta- 2 Scientists first became interested in the topic of temporal perception when they realized that there were individual differences in observers measurements of the time of stellar transits (i.e., measuring the time at which a star crossed the hairline of the telescope s eye-piece; Cairney, 1975b). Traditionally, stellar transits were measured by the eye and ear method ; That is, an observer would judge the time of the visual event (the transit) with respect to the sound of the second hand of the clock ticking in the background. Early on, differences in the focus of an observer s attention were posited as a contributory factor to these individual differences in arrival time, captured by the notion of the personal equation (see Mollon & Perkins, 1996; Scharlau, 2007). While, at first, researchers attempted to investigate the factors modulating multisensory temporal perception using variants of the complication situation (e.g., Dunlap, 1910, 1917), problems of interpretation (associated with participants eye movements to the moving stimulus etc.; Cairney, 1975b) soon led to the development of the TOJ task, and thence to studies of prior entry.

5 C. Spence, C. Parise / Consciousness and Cognition xxx (2009) xxx xxx 5 Fig. 1. Graph highlighting the prior-entry effect reported by Spence, Shore, and Klein (2001; Experiment 2) in their study of the effects of endogenously attending to a sensory modality (either vision or touch). The mean PSS is presented for the various attention conditions: Attend touch, divided attention, and attend vision. The proportions of different trial types presented in the various blocks of trials are shown to the left of each bar on the bar graph: VT crossmodal visual-tactile TOJ trial; VV unimodal visual TOJ trial; TT unimodal tactile TOJ trial. The results highlight the shift in the PSS resulting from participants endogenously shifting the focus of their attention between the tactile and visual modalities, or else dividing their attention between the two modalities. Error bars indicate the between-participant standard error of the means. neous (i.e., as indicated by participants making each response equally often) when participants attention had been endogenously directed toward the tactile modality than when it had been directed toward the visual modality instead (see Fig. 1). Using a similar methodology, Zampini and his co-workers subsequently demonstrated a significant prior-entry between vision and audition (Zampini, Shore et al., 2005), and between vision and nociception (in the form of laser painful heat stimuli; Zampini et al., 2007). Taken together, the available evidence would therefore appear to support the conclusion that prior-entry effects can be observed no matter which modality participants endogenously direct their attention toward. One generalization that can be drawn from the various studies of endogenously attending to a particular sensory modality that have been published to date (see Table 1) is that the largest prior-entry effects tend to be reported in those studies that required their participants to make TOJs while much smaller prior-entry effects of all have been observed in those studies where a SJ task has been used (see also Schneider & Bavelier, 2003; Shore et al., 2001; Stelmach & Herdman, 1991; Van der Burg, Olivers, Bronkhorst, & Theeuwes, 2008; Wada, 2003; Zampini, Shore et al., 2005; Zhuang & Papathomas, 2009). The fact that the magnitude of the prior-entry effect appears to depend on the nature of the task that participants have to perform has led certain researchers (e.g., Schneider & Bavelier, 2003; Zampini, Shore et al., 2005) to argue that response bias may only be ruled out completely by the use of an SJ task. The underlying assumption here seems to be that the fact that larger prior-entry effects have been reported in orthogonal TOJ studies than in SJ studies must reflect some form of residual response bias affecting participants performance in the former (but not in the latter) case, since many researchers appear to believe that the two tasks are thought to measure the same thing (e.g., see Schneider & Bavelier, 2003; see also Zhuang & Papathomas, 2009). However, the influence of response bias can be effectively eliminated from one s estimates of the magnitude of the prior-entry effect (no matter whether one is using an orthogonal or non-orthogonal TOJ design) by averaging participants performance on Which came first? and Which came second? versions of the TOJ task (see Shore et al., 2001). The logic here being that any response should have an equal and opposite effect on participants performance in the two cases. Hence, when performance in the Which came first? and

6 6 C. Spence, C. Parise / Consciousness and Cognition xxx (2009) xxx xxx Which came second? TOJ tasks is averaged, any response bias effects that may be present should be cancelled out, leaving just any residual attentional prior-entry effect (see below for a fuller discussion of the findings from Shore et al. s study). Finally, it should be noted that some researchers are not so convinced that the TOJ and SJ tasks do, in fact, measure the same thing (see above). 4. Prior entry resulting from attention being directed to a spatial location In recent years, the focus of much of the prior-entry research has shifted (away from the study of the effects of attending to a particular sensory modality) toward assessing the effects of attending to a particular spatial location on the perception of temporal order and synchrony/asynchrony. Research now shows that both exogenous and endogenous spatial attentional orienting can give rise to significant prior-entry effects (e.g., Shore et al., 2001; Yates & Nicholls, 2009; though see also Schneider and Bavelier (2003)). While the majority of this research has focused on the effects/consequences of the spatial orienting of a person s visual attention (e.g., Abrams & Law, 2000; Enns, Brehaut, & Shore, 1999; Hikosaka, Miyauchi, & Shimojo, 1993; Jaśkowski, 1993; Neumann, Esselmann, & Klotz, 1993; Scharlau, 2002; Scharlau & Ansorge, 2003; Scharlau, Ansorge, & Horstmann, 2006; Scharlau & Neumann, 2003a, 2003b; Schneider & Bavelier, 2003; Shore et al., 2001; Stelmach, Campsall, & Herdman, 1997; Stelmach & Herdman, 1991; Zackon, Casson, Stelmach, Faubert, & Racette, 1997; Zackon, Casson, Zafar, Stelmach, & Racette, 1999), similar results have also been reported in studies of unimodal auditory (Kanai, Ikeda, & Tayama, 2007) and tactile spatial attention (Yates & Nicholls, 2009), as well as in studies that have involved participants making crossmodal TOJs (Spence, Shore, & Klein, 2001). Unfortunately, however, the majority of these studies (especially the earlier studies) involved non-orthogonal TOJ designs. That said, orthogonal-responding TOJ paradigms (e.g., Shore et al., 2001; Yates & Nicholls, 2009), and SJ tasks (see Schneider & Bavelier, 2003) are now being used more frequently. As noted already, Shore et al. (2001) attempted to eliminate any residual response bias effects from their study of the effects of exogenous and endogenous visual spatial orienting on the prior-entry effect by averaging the performance of their participants on Which came first? and Which came second? orthogonal TOJ tasks. The participants in this study judged the order of presentation of two visual stimuli (a horizontal and a vertical line) presented from either the same or opposite sides of fixation (note that the two line segments formed a cross when they were presented from the same side of fixation). Participants attention was directed to one or other side by means of the transitory brightening of the outline of a square presented on either the left or right (the exogenous cue marked the location where the target might occur) or a central (endogenous) arrow precue at the start of each trial. The target stimuli in this experiment were just as likely to be presented on the same side as the exogenous peripheral cue as on the opposite side, and from the left or right. Importantly, however, the participants were given an incentive to direct their endogenous spatial attention in the direction indicated by the central arrow cue: in particular, the two visual TOJ stimuli (on cued trials) were more likely to appear at the cued location than they were to appear at the opposite location. Shore et al. s (2001) results showed that exogenous visual orienting led to a significantly larger prior-entry effect than did endogenous orienting (mean prior-entry effects of 61 and 17 ms, respectively; see also Jaśkowski (1993), Stelmach and Herdman (1991); though see Prinzmetal, McCool et al. (2005)). Interestingly, the residual effect of response bias in Shore et al. s (2001) orthogonal TOJ design (calculated as half the difference between the prior-entry effects reported in the Which came first? and Which came second? TOJ tasks) was estimated at 13 ms. Note that this value is much smaller than the 60 ms response bias effect observed in previous prior-entry research using a non-orthogonal TOJ design (see Frey, 1990). In other words, it would appear that the use of an orthogonal TOJ design successfully removed most of the response bias present in non-orthogonal TOJ tasks. Yates and Nicholls (2009) have recently reported a similar pattern of results in their study of the effects of exogenous and endogenous orienting of spatial attention on tactile TOJs. They reported a mean endogenous priorentry effect of 24 ms and a mean exogenous prior-entry effect of 16.5 ms in their TOJ studies. Shore et al. therefore concluded that attending to a spatial location can speed-up the relative arrival time of stimuli that are subsequently presented at that location. Schneider and Bavelier s (2003) study of unimodal visual spatial prior entry, however, came to a rather different conclusion regarding the underlying causes of the prior-entry effect. They conducted four experiments designed to investigate the consequences of the peripheral exogenous spatial cuing (using single or multiple visual cues), central cuing (elicited by the presentation of an arrow at fixation), and gaze cuing (the gaze-deviated faces were again presented at fixation) of spatial attention on temporal perception. The participants in each experiment responded by making a TOJ regarding which visual stimulus (one red and the other green) had been presented first or, in another session (conducted on a different day), judging whether the two visual stimuli had been presented simultaneously or not. Schneider and Bavelier reported that the effect of exogenous spatial attentional orienting was very much reduced in the SJ task as compared to the orthogonal TOJ task. Nevertheless, significant prior-entry effects were reported in both tasks. The exogenous spatial cuing effect peaked at a cue-target SOA of 75 ms in the TOJ task and at an SOA of 40 ms in the SJ task, but was also pronounced at the 125 ms SOA in both tasks (note that they tested SOAs of 0, 40, 75, 125, 200, 500, and 1000 ms). By contrast, while the presentation of the central arrow cue resulted in a prior-entry effect in the TOJ task at the majority of non-zero SOAs (SOAs of 0, 100, 300, 600, 1000, and 1500 ms were tested), peaking at an SOA of 600 ms, analysis of the data from the SJ task only provided evidence of a significant prior-entry effect at the 600 ms SOA in the SJ task. Gaze-cuing led to a similar pattern of results although with a slightly different timecourse.

7 C. Spence, C. Parise / Consciousness and Cognition xxx (2009) xxx xxx 7 Fig. 2. Schneider and Bavelier (2003) outlined three different mechanisms that might result in behavioral performance in psychophysical temporal judgment tasks that might appear to provide evidence in support of the prior entry hypothesis. According to Titchener (1908), attention might speed-up the transmission of one stimulus relative to the other; either by reducing the transmission time of the attended stimulus or slowing the transmission of the unattended stimulus (see Spence, Shore, & Klein, 2001; 1 in figure). Alternatively, however, results consistent with the prior-entry effect might be cognitively mediated, that is, might alter the criteria within the decision process that compares the arrival times of the two stimuli and hence lead to a behavioral outcome indistinguishable from an attentional effect (2 in figure). Finally, in those studies in which attention is oriented by means of the presentation of a peripheral spatial cue, Schneider and Bavelier suggested that local sensory interactions between the cue and target stimuli (on the valid trials, where the cue and target are presented from the same location) might result in sensory facilitation which again could be confused with an attentional prior-entry effect (see 3 in diagram). Note that Schneider and Bavelier attempted to draw a distinction between sensory enhancement (or facilitation) and attention, arguing that they reflected separate processes. They also described the effects occurring at the level of decisional processes as attentional in nature (see their Fig. 1). However, it is important to note that many other researchers (ourselves included) describe effects on decisional processes in terms of response bias (rather than attention), while considering both attentional and sensory facilitation to represent different forms of attentional effect (i.e., different kinds of early sensory enhancement resulting from either the voluntary or automatic prioritization of certain of the information available at a given moment; e.g., see Naghavi & Nyberg, 2005, p. 392). For the case of unimodal spatial cuing this facilitation was suggested to result from the activation of the same pool of receptors by both the cue and first target stimulus, whereas in the case of crossmodal cuing studies sensory facilitation was attributed by Schneider and Bavelier to temporal ventriloquism effects instead (see Morein-Zamir et al., 2003). [Adapted, modified, and redrawn from Schneider & Bavelier, 2003, Fig. 1]. On the basis of these results, Schneider and Bavelier (2003) went onto conclude that, contrary to Titchener s (1908) claim, endogenous attention (i.e., following the presentation of the central arrow cue) does not have much of an effect on arrival times. They also put forward a non-attentional account to explain at least part of the cuing effects reported following exogenous cuing. They suggested that the prior-entry effect reported there might actually partially reflect some form of spatiallyspecific sensory interaction (in their words, sensory facilitation ) taking place between the cue and the first of the target stimuli on valid trials (i.e., on those trials where the cue and target stimulus are presented from the same side or spatial location; see Fig. 2). Note that in the case of the unimodal visual exogenous spatial cuing studies reported by Schneider and Bavelier, it is likely that the same group of neurons (in the ascending visual pathways) would have been used to process both the visual cue and the first of the visual target stimuli on the valid trials (that is, the cue and target fell within the same receptive fields (RFs) having been presented from more or less the same spatial location). Note though that the same argument cannot be applied to the results of Shore et al. s (2001) study, since the RFs at the earliest cortical projection sites would likely be too small to accommodate both the cue and target stimulus used there (in valid trials). There are, however, several reasons to question Schneider and Bavelier s (2003) claim that endogenous attention has little effect on arrival times. Note, for example, that while one of the two target stimuli would be presented from the location indicated by the central arrow cue (there were eight possible target locations arranged around a virtual circle centered on fixation in their Experiment 2), there was actually little incentive for participants to attend to the location indicated by the cue (since both the first and second target were equally likely to be presented at this location). 3 Given that there was no strategic reason for participants to attend to the location indicated by the central arrow cue, it is unfortunate that Schneider and Bavelier failed to provide any independent confirmation that the central arrow cue had actually been effective in eliciting an endogenous shift of attention (see Spence, Shore, & Klein, 2001, on this problem, which is surprisingly common in studies of prior-entry; see also Schmidt, 2000). Note that a similar problem also affects the interpretation of the results of Schneider and Bavelier s central gaze cuing study. It should also be pointed out that one assumption underlying much of Schneider and Bavelier s reasoning is that TOJs and SJs actually measure the same thing. As yet, however, there seems no convincing reason to accept this claim (see above). In an elegant recent study, Weiss and Scharlau (2009) showed that the visual spatial prior-entry effect is reduced, but importantly not eliminated (it dropped from 54 ms to 44 ms), if participants were provided with feedback about the correctness of their TOJs on a trial-by-trial basis, and rewarded for making veridical (correct) responses. This result shows that while there may be a small strategic element to the prior-entry effect under certain conditions, the majority of the effect (at least in their study) reflects the obligatory consequences of attentional prioritization. 3 Our thanks to Ray Klein, for pointing out this possibility.

8 8 C. Spence, C. Parise / Consciousness and Cognition xxx (2009) xxx xxx 5. Prior-entry effects resulting from crossmodal exogenous spatial orienting Prior-entry effects have also been reported in a number of crossmodal exogenous spatial cuing studies (Eskes, Klein, Dove, Coolican, & Shore, 2007; Hongoh, Kita, & Soeta, 2008; Lupiánez, Baddeley, & Spence, 1999; Santangelo & Spence, 2009; Shimojo, Miyauchi, & Hikosaka, 1997; Spence & Lupiánez, 1998; Van Damme, Gallace, Spence, Crombez, & Moseley, 2009; Wada, 2003). For instance, the participants in an experiment by Spence and Lupiáñez were presented with a spatially-non-predictive visual cue on either the left or right at the start of each trial. The participants then had to make an unspeeded TOJ concerning which of two tones, one high-pitched and the other low pitched had been presented first (or second). One tone was presented on either side of fixation at SOAs of ms (the SOA was varied using the method of constant stimuli). The onset of the spatially-non-predictive visual cue occurred 150 ms before the onset of the first tone. In spite of the spatially-non-predictive nature of the visual cue, a small but significant prior-entry effect was nevertheless still observed (M = 40 ms) no matter whether participants performed a Which stimulus came first? or a Which stimulus came second? version of the task, thus eliminating response bias as a potential contributing factor to this estimate of the prior-entry effect. In another study, Spence and his colleagues (Lupiánez et al., 1999; Spence & Lupiánez, 1998) presented a spatially-nonpredictive auditory cue (a brief white-noise burst) on either the left or right prior to a pair of visual and tactile stimuli, again one presented on either side of fixation (see Fig. 3). The participants now had to make an orthogonal TOJ regarding which modality had been presented first (or second), regardless of the side on which the first stimulus had been presented. Even Fig. 3. Schematic illustration of a typical trial in one of the crossmodal TOJ studies conducted by Lupiánez et al. (1999). At the start of each trial, a spatiallynon-predictive auditory cue was briefly and unpredictably presented from a loudspeaker cone placed by the participant s left or right hand. This was followed by the presentation of a pair of visual and tactile stimuli, one presented to either side of central fixation. These TOJ stimuli were separated by an SOA of ms (varied using the method of constant stimuli). Participants made an unspeeded discrimination response regarding which modality (vision or touch) appeared to have been presented first (or second), while trying to ignore the auditory cue. (b) Mean proportion of vision-first (or visionsecond) responses as a function of the stimulus onset asynchrony (SOA) between the visual and tactile stimuli. Trials on which the visual stimulus was presented on the side of the auditory cue are shown by dotted line, while trials on which the tactile stimulus was presented on the auditorily-cue side are shown by solid line. The visual stimulus had to lead by a greater interval for the PSS to be achieved when the side of the tactile stimulus was cued than when the side of the visual stimulus was cued instead. The mean prior-entry effect (i.e., averaging over the Which came 1st? and Which came 2nd? resulting from the crossmodal exogenous orienting of spatial attention reported in this study was 141 ms.

9 C. Spence, C. Parise / Consciousness and Cognition xxx (2009) xxx xxx 9 though the participants were instructed to ignore the spatially-uninformative sound as much as possible, the results nevertheless once again showed that the stimulus subsequently presented on the side of the task-irrelevant cue sound was speeded-up relative to the stimulus that happened to be presented on the other side. A mean prior-entry effect of 121 ms was observed in the Which came first? study as compared to a mean prior-entry effect of 161 ms in the Which came second? version of the task (i.e., once again supporting the claim that the orthogonal nature of their TOJ task had effectively eliminated the response bias contribution to the prior-entry effect). Several researchers have also documented a crossmodal spatial cuing effect on unimodal visual TOJs resulting from the presentation of a spatially-non-predictive auditory cue on either the left or right (see Hongoh et al., 2008; Santangelo & Spence, 2009; Shimojo et al., 1997). Taken together, the results of the studies reported in this section therefore suggest that prior entry can be elicited by the exogenous orienting of spatial attention that follows the peripheral presentation of a spatially-non-predictive visual, auditory, or tactile cue. Once again, however, Schneider and Bavelier (2003) have suggested that sensory facilitation effects might help to explain the prior-entry effects reported in crossmodal exogenous spatial cuing studies. However, given that different sets of sensory receptors will, of necessity, initially be used to process the cue and target stimuli in such crossmodal studies (i.e., where the cue and target are presented to different sensory modalities), an alternative mechanism for sensory facilitation was needed to that proposed to account for intramodal exogenous spatial cuing effects (see above). Here, Schneider and Bavelier turned to the phenomenon of temporal ventriloquism (see Morein-Zamir, Soto-Faraco, & Kingstone, 2003). Temporal ventriloquism is the term used to describe the temporal equivalent of the well-known spatial ventriloquism effect (e.g., Alais & Burr, 2004); the idea is that when auditory and visual stimuli occur slightly asynchronously, the brain may attempt to bind the two stimuli together by pulling one stimulus into temporal alignment with the other. Schneider and Bavelier suggested that temporal ventriloquism might serve to pull the first target earlier in time (i.e., toward the time when the cue was presented) on the validly-cued trials (i.e., when the cue and first target stimulus were presented from the same spatial location). The important point to note here though is that subsequent research by Vroomen and Keetels (2006) and Keetels and Vroomen (2008) has shown that there is no spatial modulation of the temporal ventriloquism effect. That is, the stimuli in one sensory modality have just as much effect on the temporal perception of stimuli presented in the other modality regardless of whether they are presented from the same from versus from different locations. Hence, it turns out that the phenomenon of temporal ventriloquism simply cannot be used to explain the spatially-specific cuing effects seen in crossmodal exogenous cuing studies of the prior-entry effect. It remains an important question for future research to determine whether a plausible non-attentional (i.e., sensory facilitation) alternative explanation for these crossmodal exogenous prior-entry effects can be developed (see Rowland, Quessy, Stanford, & Stein, 2007). Nevertheless, in the meantime, the possibility that sensory facilitation effects might influence performance should certainly be borne in mind whenever one is evaluating the prior-entry effects reported in exogenous spatial cuing studies (especially in unimodal cuing studies). Van Damme et al. (2009) recently conducted a study in which the participants had to make either unimodal TOJs concerning pairs of either tactile or auditory stimuli (consisting of brief vibrotactile stimuli or pure tone bursts), one presented on either side. Shortly before the onset of the first target (at a cue-target SOA of 250 ms), a picture was presented on one or other side for 150 ms. The picture consisted of one of three different image types: an image of physical threat (e.g., the picture of a pit-bull dog intending to bite), a general threat image (e.g., the picture of a jet exploding), or a neutral picture (e.g., the picture of a boat; see also West et al., in press). The visual cue was non-predictive with regard to the side on which the first auditory or tactile stimulus would be presented. The results showed a larger crossmodal exogenous spatial prior-entry effect following the presentation of body-specific threat pictures when the participants made tactile TOJs, while the general threat pictures resulted in the largest prior-entry effect for the auditory stimulus presented on the side of the picture instead. These results therefore suggest that the exogenous prior-entry effect elicited by the presentation of a visual cue can have differential effects depending upon the precise meaning of the cue. While the results of this study are intriguing, it is also important to note that the participants task was not orthogonal (i.e., the cue was presented on the left or right and participants had to report the side of the 1st auditory or tactile stimulus). Second, the blocked nature of the experimental design (with auditory and tactile targets being presented in separate blocks of experimental trials) means that any uncontrolled changes in participants attentional control settings may have had some influence on the pattern of results that were observed (cf. Folk, Remington, & Johnston, 1992). Thus, ideally this experiment should be repeated using an SJ task with the target modality varying on a trial-by-trial basis. 6. The cognitive neuroscience of prior entry Having reviewed the behavioral manifestation of prior entry, we turn now to the putative neural substrates underlying the effect. Neuroscience evidence demonstrating the modulation of early responses in sensory cortical areas would clearly provide strong support for the genuinely perceptual nature of prior entry. It would also help to rule out alternative accounts in terms of decisional-level effects (e.g., Pashler, 1998; Schneider & Bavelier, 2003). Several studies published over the last 15 years have suggested that covert spatial attention can indeed speed-up the processing of visual information (see Carrasco & McElree, 2001; Di Russo & Spinelli, 1999; Schuller & Rossion, 2001; Spinelli, Burr, & Morrone, 1994). However, the first study to explicitly look at the attentional modulation of TOJs using ERPs failed to demonstrate any effect of spatial attention on the timing of ERPs in visual cortex (McDonald et al., 2005). The participants in McDonald et al. s study had to judge which of two visual stimuli, one red and the other green, had been presented first. One visual stimulus was presented from either