Distracted and confused?: Selective attention under load Nilli Lavie Department of Psychology and Institute of Cognitive Neuroscience, University College, Gower Street, London WC1E 6BT, UK The ability to remain focused on goal-relevant stimuli in the presence of potentially interfering distractors is crucial for any coherent cognitive function. However, simply instructing people to ignore goal-irrelevant stimuli is not sufficient for preventing their processing. Recent research reveals that distractor processing depends critically on the level and type of load involved in the processing of goal-relevant information. Whereas high perceptual load can eliminate distractor proces- sing, high load on ���frontal��� cognitive control processes increases distractor processing. These findings provide a resolution to the long-standing early and late selection debate within a load theory of attention that accommo- dates behavioural and neuroimaging data within a framework that integrates attention research with executive function. Introduction The ability to remain focused on a task is vital for any coherent cognitive function, especially when there might be potential interference from distractors that are irrele- vant for the task. However, people are often distracted by task-irrelevant stimuli. Daily life provides numerous examples: a fly hovering about might distract you while reading this article, an attractive bill-board can distract a driver, and so forth. In the laboratory, research that looked at the extent to which distractor processing can be prevented led to an enduring controversy. Mixed results as to whether focusing attention on task-relevant stimuli can exclude distractors from early perceptual processing (an ���early��� selection effect) or can only prevent distractors from controlling behaviour and memory (a ���late��� selection effect) has fuelled a longstanding debate between early- and late-selection views of attention [1]. Recent research on the role of load in the processing of task-relevant information in determining the processing of task-irrelevant distractors offers a possible resolution. This research indicates that distractor perception can be prevented (early selection) when processing of task- relevant stimuli involves high perceptual load, and that although distractors are perceived in tasks of low perceptual load (late selection), their impact on behaviour depends on other types of load, such as that on working memory. These results have therefore provided better understanding of the circumstances under which people can achieve coherent goal-focused behavior with minimal intrusions of goal-irrelevant information. Perceptual load studies: behavioural experiments Research on the role of perceptual load in selective attention was triggered by the hypothesis that perception has limited capacity (as in early-selection views) but processes all stimuli in an automatic mandatory fashion (as in late-selection views) until it runs out of capacity [2,3]. This led to the predictions that high perceptual load that engages full capacity in relevant processing would leave no spare capacity for perception of task-irrelevant stimuli. In situations of low perceptual load, however, any capacity not taken up in perception of task-relevant stimuli would involuntarily ���spill over��� to the perception of task-irrelevant distractors. These predictions were tested in experiments that assessed the effects on distractor perception of varying perceptual load in the task-relevant processing [3���5]. Increased perceptual load means that either the number of different-identity items that need to be perceived is increased, or that for the same number of items perceptual identification is more demanding on attention [3���5] (see Figure 1). These experiments found that increased perceptual load reduces, indeed typically eliminates, any distractor interference effects, in support of the perceptual load hypothesis. Reduced distractor interference under conditions of high perceptual load is not simply the result of the general increase in task difficulty with load and the associated slowing of performance. Manipulations of extreme sensory degradation (e.g. reducing the target size or contrast so much so that it is barely seen) that cannot be compensated for by applying more attention ��� in other words subjecting target identification to sensory ���data limits��� rather than attentional ���resource limits��� [6] ��� increase the general task difficulty (i.e. reduce speed and accuracy, compared with an intact target) but do not reduce distractor interference [7]. Alternative accounts to perceptual load in terms of general task difficulty or slowing are also ruled out by the findings (reviewed later) that increasing load on cognitive control processes (e.g. working memory) increases task difficulty but has the opposite effect to perceptual load, resulting in an increase (rather than a decrease) in distractor interference. The studies mentioned so far assessed perception of the distractor identity in the ���response competition��� paradigm (Figure 1). Other paradigms used since have included Corresponding author: Lavie, N. (n.lavie@ucl.ac.uk). Available online 5 January 2005 Review TRENDS in Cognitive Sciences Vol.9 No.2 February 2005 www.sciencedirect.com 1364-6613/$ - see front matter Q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.tics.2004.12.004

measures of implicit learning about spatial configuration of irrelevant distractors [8], as well as measures of both positive priming (Thoma and Lavie, unpublished), and negative priming effects from distractors that are pre- sented as targets on subsequent trials [9]. All these different approaches have converged in showing that distractor effects are eliminated under high perceptual load in target processing (but see Box 1 for some exceptional distractors). This generalization of percep- tual-load effects across multiple measures of distractor processing provides support for the suggestion that distractors are simply not perceived when the perception of task-relevant stimuli under high load consumes all or most of the available capacity. Neuroimaging studies also indicate this, as reviewed next. Effects of perceptual load on distractor processing in the brain Several neuroimaging studies show that high perceptual load in a relevant task modulates neural activity related to irrelevant distractors. In one study [10] neural activity in visual cortex associated with the perception of irrelevant motion distractors was determined by the level of load in a relevant task performed on words at fixation. Subjects were asked either to monitor a word���s case (low load) or number of syllables (high load). Irrelevant motion back- ground evoked responses in motion selective cortices (e.g. MT, V1/V2) in low-load but not high-load conditions. In another study [11], activations related to written words were not elicited when subjects ignored them while performing a high-load task of monitoring a rapid TRENDS in Cognitive Sciences 00 Go (a) (b) X N X N O O O O O X N K N Z X M H W (d) 4 X 7 N X * * * * * * X 7 N X 40 4 0 10 20 30 40 50 60 70 39 61 0 10 20 30 40 50 60 70 (c) (e) Perceptual load Working-memory load RT incongruent ��� congruent (ms) RT incongruent ��� congruent (ms) No Go Low High Low High 147326 * * * * * * Figure 1. Examples of stimuli and results in behavioural load experiments, using the response-competition paradigm. (a���c) Subjects make speeded responses indicating whether a central target letter is one of two pre-specified letters (X or N) while attempting to ignore a peripheral distractor letter. Slower responses in the presence of an incongruent distractor (shown in a) compared with a congruent distractor (e.g. X distractor for X target) indicate that the distractor identity was perceived. (a) Perceptual load is manipulated by varying the number of items (letters) that are similar to the target (no similar items in low load, left five in high load, right) [4]. Other experiments [3,5] varied the number of task-relevant items by presenting the target letter with fewer (up to three) non-target letters of different identities in conditions of low perceptual load. (b) Perceptual load is manipulated by increasing perceptual processing requirements for the same displays. Whether target letter responses are made (Go trials) or not (No-Go trials) depends either on detecting the presence of any blue shape (low load), or on discrimination of conjunctions of colour and shape (high load e.g. target responses are made only if there is a blue square and a red circle). See [51] for review of previous evidence that feature versus conjunction tasks impose low and high load, respectively, on attention. (c) Distractor effects are greater in low than in high perceptual load conditions. (d) Working-memory load is manipulated during performance of a response-competition task. Subjects are required to memorize the set of either one (low load) or six (high load) digits presented at the start of each trial, to indicate whether a memory probe digit presented at the end of each trial was present or absent in the set. (e) Working-memory load has the opposite effect on distractor processing to perceptual load: distractor effects are greater in high than in low working-memory load. Review TRENDS in Cognitive Sciences Vol.9 No.2 February 2005 76 www.sciencedirect.com