E-Book, Englisch, 413 Seiten
Folk / Gibson Attraction, Distraction and Action
1. Auflage 2001
ISBN: 978-0-08-049955-0
Verlag: Elsevier Science & Techn.
Format: EPUB
Kopierschutz: 6 - ePub Watermark
Multiple Perspectives on Attentional Capture
E-Book, Englisch, 413 Seiten
ISBN: 978-0-08-049955-0
Verlag: Elsevier Science & Techn.
Format: EPUB
Kopierschutz: 6 - ePub Watermark
Attraction, Distraction and Action
Autoren/Hrsg.
Weitere Infos & Material
2 Inhibition of Return in Monkey and Man
Raymond M. Klein; Douglas P. Munoz; Michael. C. Dorris; Tracy L. Taylor Introduction
The capture of attention by the objects of our experience can be explored at a phenomenological, behavioral or neural level of analysis. Although, alone, each of these approaches is limited, some of these limitations can be overcome by collecting data and developing explanatory frameworks that link the different levels of analysis (Teller, 1990). We are excited by the fruitfulness of an interdisciplinary approach in which behavioral/cognitive analysis is combined with the collection of neuroscientific evidence that is aimed at determining how the mechanisms which affect behavior are implemented in the nervous system (cf Trappenberg et al., 2001). In this chapter we will describe a collaborative effort to explore just one mechanism pertinent to the orienting of visual attention, inhibition of return (IOR, see Klein, 2000, for a review). We hope that our interdisciplinary strategy and initial successes will serve as a model or blueprint for similar collaborations on other topics covered in this volume. Valuable information for a cognitive-neuroscientific understanding of any mind-brain relation can be provided by studying the breakdowns in human behavior that follow from damage due to insult, stroke or disease. The same could be said for modem neuroimaging methods which can provide temporally (magnetic encephalography; event-related potentials) or spatially (functional magnetic resonance imaging) precise information about the net activity level of many neurons in a brain region. Valuable information about IOR has been provided using both of these approaches (Hopfinger & Mangun, 1998; McDonald, Ward, & Keihl, 1999; Posner et al., 1985; Sapir, Soroker, Berger, & Henik, 1999). We believe that for a cognitive analysis aimed at how cognitive computations are implemented in the neural machinery (Marr, 1982) an analysis of the activity level of single neurons is also essential. Whereas the technique of single unit recording in the alert nonhuman primate is well-matched for this purpose, its implementation and connection with studies of human performance requires a task that both humans and monkeys can perform and whose mechanism of interest is manifest and similar in both species. Model task: IOR in a cue-target paradigm
Although it has since been explored using more complex displays and task requirements, IOR was first observed using a highly simplified, “model” task. In Posner and Cohen’s (1984) initial demonstration, the observer was presented with a fixation display consisting of 3 horizontally arranged boxes. One of the peripheral boxes brightened briefly and, at varying intervals after the onset of this “cue”, a target was presented. Observers made simple manual detection responses to peripheral targets that appeared within the left or right box with equal probability (i.e., regardless of the location of the cue). As had been observed in previous studies, at short cue-target intervals, targets were detected more rapidly at the cued than at the uncued location. This finding is typically interpreted as a processing benefit due to the reflexive allocation of attention to the luminance increase (the cue) in the periphery.1 Posner and Cohen used two methods to encourage observers to remove their attention from the cued location. Either there was a brightening of the fixation box (shortly after the onset of the peripheral cue) or central targets were presented with a high probability (60%, compared to 10% on each side). Because peripheral targets were equally likely to occur at the cued and uncued locations, one might naturally expect that after the initial facilitatory effects of the cue had waned, performance at the cued and uncued locations would become similar. Instead, beginning about 250 ms after cue onset, there was a cross-over interaction wherein performance became worse at the cued location for the remaining intervals (up to 500 ms in this experiment). This inhibitory aftereffect of the cue was subsequently named “inhibition of return” (Posner et al., 1985) and since Posner’s seminal experiments we have learned a great deal about the timecourse, spatial distribution, causes, effects and development of IOR (for reviews, see Hood, Atkinson, & Braddick, 1998; Klein, 2000; Taylor & Klein, 1998). The activation of oculomotor activity, in the form of a saccadic program that is executed or merely planned, seems to be a pre-requisite for IOR. Typically, a peripheral event is used to cause IOR, and such an event will activate a tendency to foveate it (whether or not a saccade is executed). It was Rafal et al.’s (1989) finding of IOR following the cancellation of an endogenously prepared saccade (in response to an arrow cue at fixation) but not following the cancellation of an endogenously generated shift of attention that definitively linked the cause of IOR with oculomotor programming. Once caused, IOR affects both manual and saccadic responses to peripheral targets, delaying responses to targets presented in the vicinity of the previous cue. If the eyes move after a cue but before a target is presented, it is the location in space defined by environmental coordinates -- not by retinal coordinates -- that is inhibited. However, if an object is cued, and then the object moves through space before a target is presented, some if not all of the inhibition moves with the object (Abrams & Dobkin, 1994a; Tipper et al., 1991; Tipper et al., 1997; Tipper et al., 1999). Thus, IOR is also coded in an object-centered frame of reference. IOR is linked to the concept of attentional capture in two ways. First, under some views, IOR is conceptualized as the aftermath of attentional capture (i.e., as resulting from the disengagement of attention from the captivating stimulus). Second, the presence of IOR modulates the attention-capturing ability of subsequent targets. When the visual system is given a choice of two targets to inspect, for example, there is an increased probability of choosing the target that is not at a previously cued (Posner et al., 1985) or previously fixated (Peterson et al., 2000) location. In this sense, IOR, laid down by the previous overt or covert orienting experience, biases subsequent orienting in new directions (Klein & MacInnes, 1999). Whereas the effect of IOR is linked to output levels of processing (e.g., Ivanoff & Klein, 2001; Taylor & Klein, 2000) associated with spatial orienting, it is not restricted to motoric processes. Strongly supporting a reduction in signal amplitude, Handy et al. (1999) showed that IOR decreases d’ and Klein & Dick (in press) found that IOR decreases the probability of correct target identification using a dual-stream RSVP task designed to minimize the contribution of motoric processing. Similarly, several studies of event-related potentials (ERPs; Hopfinger & Mangun, 1998; McDonald, Ward & Keihl, 1999) have reported that IOR reduces the sensory response (magnitude of the P1, an early ERP component seen about 100 ms after stimulus onset) generated by stimuli presented at previously cued locations, thereby giving rise to a “salience” differential in favour of items at uncued locations. Functional significance
Perhaps the most important question to consider before launching an interdisciplinary attack is, “What is the function of IOR?” Suppose you are looking for something (on your desk) or someone (in your classroom). When the item you are looking for does not pop out of the visual array, one strategy you might employ is to inspect each item in turn until the target is found. Such a search might be deliberate and controlled by a pre-planned order (as when you search for typographical errors when proofreading); or it might be more haphazard, beginning with candidate items whose initially detected features indicate a potential match with the target. The latter kind of serial search strategy, which has been supported in many search experiments (though not necessarily to the exclusion of other strategies) is referred to as “guided search.” One problem confronting any serial search that is not controlled by a pre-planned sequence of inspections is how to keep track of which items have already been examined and identified as non-targets. Such a search would be increased in efficiency by a mechanism that discouraged attention or gaze from returning to previously inspected objects or regions where the goal object had not been found. As first proposed by Posner and Cohen (1984), inhibition of return (IOR) may be such a mechanism. Supporting this idea, Klein (1988; Klein & MacInnes, 1999; see also, Müller & von Mühlenen, 1999; Peterson et al., in press; Takeda & Yagi, 1999) found direct evidence that an inhibitory tagging mechanism operates during serial search, presumably as a “foraging facilitator.” Neuroscientific answers to unsolved problems
At its most fundamental level, the cognitive neuroscientific approach we have begun is aimed at discovering how the behavioral inhibition that has been observed in studies of human performance is implemented in the nervous system. As seen below, we have so far only “looked” in one neural structure known...




