Attraction, Distraction and Action

Multiple Perspectives on Attentional Capture

Charles Folk, Bradley Gibson

• Publisher: North Holland
• Year: 2001
• ISBN:
  9780080499550
• Topic: Psychology
• Subtopic: Experimental Psychology
• Index: Psychology

Over the last decade there has been a spate of research on the empirical phenomenon known as "attentional capture". Interest in capture can be attributed not only to its applied significance, but also to the implications of the phenomenon for theories of selective attention, as well as cognitive control in general. This growing interest, however, has also spawned a wide variety of experimental paradigms, empirical results, and theoretical perspectives. In June of 2000, 40 experimental psychologists converged on Villanova University for a conference and workshop on attentional capture. The intent was to provide an intimate forum for scientists from diverse perspectives and backgrounds, and using diverse methodologies to present their research on attentional capture and also engage in small group discussions on such key issues as the definition, measurement, and theoretical treatment of attention capture. This book presents a collection of chapters based on those presentations and discussions. Chapters are organized around areas such as neuroscience, visual cognition, developmental, individual differences and dynamical systems. The volume provides: a summary of the latest cutting edge research; an important compass for future research in this area; a useful survey of the field; contributions from internationally recognized experts in attention. Due to its exclusive focus on the topic of attentional capture the volume should make an excellent supplemental text or reference book for advanced undergraduate or graduate seminars in cognitive psychology and attention.

Part I

Neuroscience

1

Electrophysiological Studies of Reflexive Attention

Joseph B. Hopfinger; George R. Mangun

Abstract

Models of human cognition hold that information processing occurs in a series of stages. Cognitive psychology, in particular, is concerned with the internal mental processes that begin with the appearance of an external stimulus and result in a behavioral response. An enduring question has focused on determining the stage or stages of information processing at which attention might have an influence. Measures of overt behavior have long been used to make inferences about the internal mental mechanisms of attention. Increasingly though, physiological measures of human brain activity have been used to provide direct measures of discrete stages of information processing during attentional performance. In this chapter, we briefly review the event-related potential (ERP) approach to the study of attention, and present recent results utilizing this methodology in the study of reflexive attentional capture. These experiments have revealed that reflexive attention is able to influence multiple stages of information processing beginning at a relatively early stage of visual cortical analysis.

Background

Tracking information processing in the brain: Electrophysiological methods

The development of electrophysiological recording techniques dates back to the early 1930’s, when Hans Berger and Herbert Jasper developed techniques that would later be used to directly examine the neural mechanisms of the human brain’s attention systems (Jasper, 1935). By recording from electrodes placed on a human subject’s scalp, they were able to measure small voltage fluctuations that reflected underlying neural activity. The recording of the ongoing voltage variations measured on the scalp is known as the electroencephalogram (EEG) and is now known to be primarily a measure of the post-synaptic (dendritic) potentials from populations of synchronously active and aligned neurons (see Nunez, 1981 for a more comprehensive discussion). Early electrophysiologists analyzed large rhythmic fluctuations in the EEG (e.g., “alpha” waves) that could index overall states of arousal (e.g., Jasper, 1935). Although the ongoing EEG can provide a measure of the subject’s global brain state, it is not as well suited for identifying patterns of brain activity associated with specific types of stimulus processing or specific mental functions. This is due to the fact that the larger rhythmic potentials of the ongoing EEG may be several times larger in amplitude than the relatively small fluctuations produced by neural activity supporting individual mental events. The EEG reflects processes occurring throughout the brain related to a host of mental activities, as well as voltage fluctuation that are not due to brain activity (e.g., “artifacts” generated by muscles on the head or neck). As a result, the neural activity generated by a specific mental event of interest can be difficult or impossible to observe in the ongoing EEG.

The voltage fluctuations produced by particular events of interest can, however, be detected using signal averaging procedures. For example, the neural activity produced by a specific visual stimulus can be measured if the ongoing EEG is averaged over multiple occurrences of that specific visual event (Figure 1). Epochs of time surrounding the visual event of interest can be extracted from the EEG record and averaged together, after aligning the onset of the visual stimulus for each epoch. The voltage amplitude can then be averaged at each timepoint separately, resulting in a single event-related-potential (ERP) waveform. The ERP thus represents the response to a specific event, timelocked to the onset of that event. The averaging process effectively cancels out the electrical activity in the EEG that is not time-locked to the stimulus event of interest. This occurs because on average over many trials, the uncorrelated activity is just as likely to be of positive or negative polarity at any post-stimulus time point. Given a sufficient number of trials, the averaging process leaves only the activity evoked by the event of interest.

Figure 1 At left is shown an example of the scalp recorded electroencephalogram (EEG), recorded continuously while the event of interest (in this case a visual stimulus: S) is presented multiple times. Epochs of the EEG surrounding the onset of the visual event are extracted, aligned according to the onset time of the event of interest, and then averaged point by point (middle column). The resulting average is referred to as the Event-Related Potential (ERP; right column). Note that the amplitude of the ERP is much less than that of the EEG, a typical situation that necessitates the averaging procedure.

A canonical ERP waveform consists of a series of voltage fluctuations, representing positive and negative potentials generated by the event of interest. As shown in Figure 1, the voltage fluctuations are typically labeled according to: (1) polarity (“P”ositive, or “N”egative; note that the convention followed here plots positive voltages downward); and (2) order of occurrence (P1 = 1^st major exclusively-positive component) or latency of occurrence (e.g., the peak of the NP80 component, which can be negative or positive, depending on the location of the visual stimulus, occurs at approximately 80 ms latency). The “prestimulus” period represents the activity time-locked to the event of interest that occurs before the stimulus appears. Since the event of interest has not yet occurred, under most circumstances, there should be no systematic activity before the onset of the event of interest. Therefore, this period can be used as a measure of the effectiveness of the averaging procedure in eliminating activity that is not due to the event of interest.

Using ERPs, it is possible to measure neural activity from the moment in time a stimulus is presented, through multiple levels of processing, up to and including response execution. ERP components can be related to hypothesized stages of mental processing (indicated schematically in Figure 2). Although much work remains to be done in order to understand the specific mental functions subserved by each particular ERP component, many of these components can at least be classified as underlying simple sensory processes or higher order cognitive processes. The ability to track mental processing in real time has proven very useful in helping to elucidate the stage(s) of processing that attention may act upon to modify mental processing.

Figure 2 Shown at the top are a few of the many hypothesized stages of information processing that intercede between the initial presentation of a physical stimulus and an eventual response to that stimulus. At bottom, an ERP waveform is shown approximately aligned with the hypothesized stages of processing. The ERP waveform shown here is only for illustration purposes - the components shown here are typically observed at different scalp sites; not all would be observed at the same scalp location. In this chapter, we will focus mainly upon the sensory P1 component, and on the P300 component that indexes post-sensory higher-order cognitive processing.

Early versus late selection

A classic debate in psychology has concerned the nature of our ability to filter out unwanted information. Specifically, the debate concerns the level of information processing at which relevant information is selected. One possibility is that this selection process occurs only just before a response must be made. This would be the extreme version of the “late-selection” argument that holds that all information received by the senses is fully processed to the level of semantic meaning (e.g., Deutsch & Deutsch, 1963). Accordingly, all sensory inputs would be completely processed, and selection would involve choosing to respond to one of several completely processed inputs (e.g., Allport et al., 1985). Alternatively, as suggested by “early-selection” theories, information may get filtered out well before it is ever processed to a level of semantic meaning. Broadbent (1958) argued that selective attention acts as a gate that allows only the desired information to proceed to higher-order processing, while keeping out all irrelevant information. Treisman (1960) argued along less extreme lines that attention acts to attenuate, rather than completely filter out, the processing of unattended inputs.

Eason, Harter, and White (1969) used the ERP technique to show that alertness and attention could affect pre-decision level neuronal processing. Specifically they showed that attentional alertness could alter neural processing of a visual stimulus as quickly as 200 ms after the presentation of a stimulus. Van Voorhis and Hillyard (1977) showed that covert (in the absence of any overt eye movements) visual selective attention could enhance visual processing starting within about 100 ms after stimulus presentation. Further investigations have shown that the P1, a positive deflection in the visual evoked ERP that peaks around 90-110 ms latency and is maximal at posterior occipital scalp sites, is the earliest visually evoked component to be reliably affected by spatial attention (e.g., Clark & Hillyard, 1996; Luck, Hillyard, Mouloua, Woldorff, Clark, & Hawkins, 1994; Mangun & Hillyard, 1988; 1990, 1991). The P1 component is referred to as a visual sensory component, in that it is evoked by visual stimuli and is sensitive to physical features of the stimulus. Scalp current density mapping and dipole modeling of scalp recorded electrical activity in attention studies have suggested that these P1 attention effects, produced by voluntary spatial selective attention, are generated in lateral extrastriate cortex (Gomez Gonzalez, Clark, Fan, Luck, & Hillyard, 1994; Mangun, Hillyard, & Luck, 1993). Combined ERP and functional neuroimaging studies have provided further evidence that the P1 is generated in the fusiform gyrus of extrastriate cortex in humans (Heinze et al., 1994; Mangun, Hopfinger, Kussmaul, Fletcher, & Heinze, 1997; Woldorff et al., 1997). Many investigations, using multiple disciplines, have thereby converged on the conclusion that voluntary attention can affect neural processing at relatively early levels. However, there are components of the visual ERP that occur earlier than the P1 that are not reliably modulated by selective voluntary attention. The NP80 component, thought to be generated by activity in the striate cortex (area V1), has not been found to be reliably affected by voluntary selective spatial attention (e.g., Clark & Hillyard, 1996). Although some neuroimaging and non-human primate studies have provided evidence for attention-related modulations in the striate visual cortex (e.g., Worden & Schneider, 1996; Motter, 1993), a recent combined neuroimaging and ERP study found that the modulation of ...

Table of contents

• Cover image
• Title page
• Table of Contents
• Copyright page
• Preface
• Contributors
• Part I: Neuroscience
• Part II: Visual Cognition
• Part III: Multiple Modalities
• Part IV: Developmental
• Part V: Individual Differences
• Part VI: Dynmical Systems/Evolution
• Subject index