Time to change

The modern world is designed to capture our attention. Traffic lights, flashing phones, moving banners, and brightly coloured billboards depicting happy faces and sensual bodies – all attempt to convey a message immediately, efficiently and compellingly.

This explosion of information bombarding our senses raises fundamental questions about how effective these attention-drawing methods are. Are we at the mercy of the incoming stimulus or do we select only those signals relevant to us? To what extent does this depend on the type of stimulus, and to what extent on the cognitive state of the observer?

Attention grabbers

Intuitively, dynamic stimuli – ones that change over time – appear to be the best candidates for automatically drawing our attention. They appear to be particularly effective if this change is rapid and fairly unexpected. The sudden onset of changing traffic lights seems a good example of such an attention-drawing stimulus.

Such observations have been confirmed in the laboratory by Watson and Humphreys (1995). In one of their experiments, participants were instructed to search for a target letter among a variable number of distractor letters dispersed across the display. Normally this search takes some effort, expressed by the increase in response times with the increase in the number of distractors. In one of the crucial conditions, however, one of the line segments of the target (e.g. the horizontal line in the letter A) would appear suddenly. Now it was always rapidly and efficiently found, to the extent that the number of other letters in the display had no effect on response times whatsoever. This means that the abrupt onset must always have been one of the first attended objects.

However, the finding that observers can efficiently select dynamic stimuli does not mean that they have to. In Watson and Humphreys' (1995) experiment, participants presumably wanted to select the dynamic target, since that was their task. This means that selection may have been under considerable control of the observer. Similarly, when we are driving, we may become aware of changing traffic lights because we are actively on the look-out for relevant information (which in the case of driving is often dynamic in nature), not because the sudden onset of the light intrinsically and automatically demands our attention. We can therefore only speak of purely automatic capture of attention if the stimulus is completely irrelevant to us, but draws our attention nevertheless.

Theeuwes et al. (1998) created exactly such a condition. They presented observers with displays consisting of a number of discs, one of which had a different colour from the rest. The task was to make an eye movement to this oddly coloured disc. However, at the same time, another disc appeared abruptly in a previously unoccupied position.

The important finding was that even though this abruptly appearing object was completely irrelevant to the observers, it attracted about as many eye movements as the target itself. Following other researchers (Yantis & Jonides, 1984), Theeuwes et al. (1998) concluded that abrupt onsets truly capture attention automatically, even if it is against the observer's interest. This conclusion appears to fit well with everyday experience, for instance when – often to our annoyance – we seem unable to ignore the flashing adverts on websites, or the changing subtitles in a foreign film.

Out of control?

So it appears that, at least in the case of single abrupt onsets, we are completely at the mercy of the stimulus. But appearances can be deceptive. To be able to decisively conclude that we have no control at all over the selection of new objects, we must return to the situation where we would actually want to select the dynamic information, for instance when we expect a long-awaited bus to emerge from around the corner. Now, if selection is indeed completely stimulus-driven, then the fact that we actually want to select the new information should have no effect whatsoever. In other words, whether you're expecting the bus or not should not affect how quickly you spot it when it finally appears. In contrast, when we do have some control over the selection of dynamic information, then our behavioural goals and expectations should affect these selection processes.

This issue has been at the core of the research conducted by Derrick Watson, Glyn Humphreys and myself (see Watson et al., 2003, for an overview). In our experiments we typically ask observers to search for a particular target object among a number of distractor objects. However, the presentation of the objects is divided up in time: half the number of items is presented first; the second half follows after one second. Crucially, the target object is always in the second set. Knowing this, observers should therefore try to ignore the previewed, 'old' set of items, and prioritise the second, abruptly appearing 'new' set (just like you know that your bus always comes last).

If observers can indeed ignore the previewed half, search should be restricted to the second half only. In other words, response times should resemble those for displays in which the first half is not even there (i.e. a 'half-set' condition). In contrast, if observers completely fail to ignore the first set, then response times should resemble those for displays in which both sets are presented at the same time (i.e. a 'full-set' condition). Our typical finding is that search times in the preview condition deviated substantially from the full-set condition and resembled that in the half-set condition. Thus, observers indeed fully prioritise the new items.

So how can we assess if this prioritisation is at least partly due to the observers' anticipation, and not solely due to the automatic attentional capture mechanisms demonstrated by Theeuwes and others? As pointed out by Donk and Theeuwes (2001), why adopt the position that observers are actively trying to prioritise a stimulus if the stimulus already does it for them?

One way of demonstrating the observers' active control of selection is to occupy them with a secondary task during the preview period. The idea is that observers cannot at the same time fully comply with the additional task and the task of anticipating new objects. This is exactly what we found (Olivers & Humphreys, 2002). Detection of a target in the new set was slower when, during the preview period, observers were still completing an additional task of identifying a briefly presented target letter. If selection of abruptly appearing stimuli had been fully automatic, then no such slowing should have been found. In real life this may mean missing a changing traffic light when we have just finished an important phone call.

Another way of demonstrating attentional control over the prioritisation of dynamic stimuli is by changing the task relevance of the old and new sets. For this purpose, Olivers et al. (2002) compared a standard preview task in which the target was always in the second set (and thus the second set should be fully prioritised) with a preview task in which the target could appear either in the first or in the second set. They found that in the latter task, fewer new items were prioritised than in the standard task. This task-dependence of the efficiency with which new information is prioritised indicates that selection is at least partly under the control of the observer.

Visual marking

Now that we have established that observers can actively control the segregation of old and new information, the question arises about how this control is implemented. What we have been proposing is that, in addition to the stimulus-driven activation of the new set (Donk & Theeuwes, 2001), the old set is inhibited below baseline level so that the new items will receive an extra advantage. This inhibitory process actively 'marks off' old objects as having been seen, and has therefore been labelled visual marking (Watson & Humphreys, 1997).

Direct evidence for a visual marking mechanism comes from a study by Olivers and Humphreys (2003). It used displays in which observers were asked to look for a red vertical bar in a set of red tilted distractor bars. Among these red bars was also a unique, but completely irrelevant, green distractor bar. Under normal conditions, despite its irrelevance, such a green bar will attract attention and delay response times.

The crucial condition was one in which the red bars were again accompanied by a green distractor, but now the complete set was preceded by a preview of another set of irrelevant green bars. The idea was that if, in anticipation of the new red set (including one new green distractor), observers actively inhibit the old green set, then some of this inhibition may spread to the new green distractor. In other words, the interference effects of the new green distractor should be reduced, because green has already been suppressed during the preview period. This reduction in interference costs is exactly what we found and provides direct evidence for a visual marking mechanism.

Attention to static stimuli

So far we have seen that observers can efficiently direct their attention to dynamic stimuli. An important question that Jan Theeuwes, Yaïr Pinto and I have recently started to investigate is whether observers can also efficiently direct their attention to static stimuli. Imagine you are searching a website filled with various moving banners and blinking adverts for some item that turns out to be the only static object on the page. Sounds like a real pain, right? The work by Theeuwes et al. (1998) and Yantis and Jonides (1984) explained earlier suggested that observers have difficulty ignoring a single dynamic stimulus. So how do you ignore a whole screen of items screaming for your attention?

It turns out that this may be remarkably easy. Pinto et al. (2006) asked observers to search for a horizontal or vertical bar among tilted distractors. In one condition all bars were standing still. In the dynamic conditions all distractors were either blinking or wiggling randomly, whereas the target was the only item that remained static. The results showed that search for the static target was much easier when all the distractors blinked or wiggled frantically. Apparently, contrary to what may be expected on the basis of single dynamic items, observers can effectively ignore multiple dynamic objects.

A role for the parietal lobe

Taken together then, there is considerable evidence that whether dynamic stimuli receive priority is to a large extent under the attentional control of the observer (see also Folk et al., 1992), some of which may be inhibitory in nature. That such attentional control mechanisms are by no means trivial becomes poignantly apparent when they fail after certain brain damage (Olivers & Humphreys, 2004). We compared performance in a preview experiment for a group of patients with lesions to the posterior parietal lobe, split up for their 'good' (ipsilesional) and 'bad' (contralesional) visual fields, to an age-matched control group.

In general, the patient group was slower and more affected by distractors than the control group. However, performance suffered disproportionally in the preview condition, in which the distractor groups were separated in time. Especially in the contralesional (bad) field, patients had serious trouble in segregating the new objects (containing the target) from the old. This suggests that the parietal lobe may be especially tuned towards dynamic stimuli. A similar conclusion was reached on the basis of fMRI studies, in which it was found that the preview condition leads to an especially active superior parietal lobule (Olivers et al., 2005; Pollmann et al., 2003).

The lesion and imaging research is leading us to some exciting new hypotheses about the relationship between brain function and attention. Parietal lesions have repeatedly been associated with attention deficits, and activity in the parietal lobe has been found in numerous attention tasks. Interestingly, most of these tasks tested for attention to non-dynamic properties, such as location, colour, orientation and shape, or more complex stimuli like faces. This has led to the idea that the parietal lobe is part of a universal attention network, which actively directs selection processes to task-relevant properties, whatever these properties are.

However, our finding that the parietal lobe is especially involved in the processing of abruptly appearing new stimuli may suggest a different source of the universal involvement of the parietal lobe in attention tasks. This involvement may not be due to these tasks requiring universally attention, but due to our tendency as vision scientists to use universal stimuli. Almost without exception stimuli are presented in an instance on a computer screen, regardless of the task. Whether these are pictures, words, colours, faces, textures or meaningless shapes; they all appear abruptly. This raises the possibility that what we are actually measuring is this universal abruptness of our stimuli, rather than universal attention mechanisms operating upon them (see Gibson & Kelsey, 1998, for related ideas).

The outside world is in some respects more dynamic than the laboratory (for instance because observers themselves are allowed to move around, such as when driving), but it is also more stable, as objects tend to stay around for a while or appear and disappear more gradually (such as overtaking cars disappearing in the distance). Attention may operate in fundamentally different ways in this more plastic spatial-temporal landscape.

A challenge for the future is therefore to design clever experiments that take out the staccato nature of the way we usually present our stimuli, and measure perceptual mechanisms in a much more fluid world.

Practical implications

The discrepancy between the laboratory and the real world does not mean that there are no practical implications, especially when the real world itself is becoming increasingly designed and artificial. I have already alluded to website design. The fact that observers can efficiently ignore multiple dynamic stimuli raises questions regarding the effectiveness of too many animated adverts. Furthermore, the finding that observers, while engaged in other tasks, may fail to efficiently select what has been regarded as the most salient type of stimulus (i.e. abrupt onsets), raises important questions about the increasing number of demands and distractions directed to drivers (such as phone calls, navigation systems, DVD players).

On the other hand, the fact that observers can make use of the dynamic aspects of stimuli to segregate relevant from irrelevant information opens up a range of possibilities for efficiently conveying important messages. One may think of route signs that give directions in stages, or intelligence maps that are built up in temporally segregated layers. No doubt there are dynamic times ahead of us.

- Chris Olivers is Assistant Professor in the Department of Cognitive Psychology at Vrije Universiteit Amsterdam. E-mail: [email protected].

Weblinks

Chris Olivers' homepage: olivers.cogpsy.nl.
Blinking, flashing and temporal response: colorusage.arc.nasa.gov/flashing.php.

Discuss and debate

Is there a limit to how many dynamic stimuli can receive priority at a given moment? Yantis and colleagues have suggested up to four, but visual marking studies between 10 and 20. How should these findings be reconciled?
Some research has suggested that even single dynamic stimuli can be ignored as long as the observer strongly concentrates on a specific location (Theeuwes, 1991; Yantis & Jonides, 1990). Is attention to locations special, in that it can override stimulus-driven attentional capture?
How can we make laboratory tasks more realistic without making them less well controlled?

References

Donk, M. & Theeuwes, J. (2001). Visual marking beside the mark: Prioritizing selection by abrupt onsets. Perception and Psychophysics, 63, 891–900.
Folk, C., Remington, R.W. & Johnston, J.C. (1992). Involuntary covert orienting is contingent on attentional control settings. Journal of Experimental Psychology: Human Perception and Performance, 18, 1030–1044.
Gibson, B.S. & Kelsey, E.M. (1998). Stimulus-driven attentional capture is contingent on attentional set for displaywide visual features. Perception and Psychophysics, 24, 699–706.
Olivers, C.N.L. & Humphreys, G.W. (2002). When visual marking meets the attentional blink: More evidence for top-down, limited capacity inhibition. Journal of Experimental Psychology: Human Perception and Performance, 28, 22–42.
Olivers, C.N.L. & Humphreys, G.W. (2003). Visual marking inhibits singleton capture. Cognitive Psychology, 47, 1–42.
Olivers, C.N.L. & Humphreys, G.W. (2004). Spatiotemporal segregation in visual search: Evidence from parietal lesions. Journal of Experimental Psychology: Human Perception and Performance, 30, 667–688.
Olivers, C.N.L., Humphreys, G.W., Heinke, D. & Cooper, A.C.G. (2002). Prioritization in visual search: Visual marking is not dependent on a mnemonic search. Perception and Psychophysics, 64, 540–560.
Olivers, C.N.L., Smith, S., Matthews, P. & Humphreys, G.W. (2005). Prioritizing new over old: An fMRI study of the preview search task. Human Brain Mapping, 24(1), 69–78.
Pinto, Y., Olivers, C.N.L. & Theeuwes, J. (2006). When is search for a static target among static distractors efficient? Journal of Experimental Psychology: Human Perception and Performance, 32, 59–72.
Pollmann, S., Weidner, R., Humphreys, G.W., Olivers, C.N.L., Müller, K., Lohmann, G., et al. (2003). Separating segmentation and target detection in posterior parietal cortex – An event-related fMRI study of visual marking. NeuroImage, 18, 310–323.
Theeuwes, J. (1991). Exogenous and endogenous control of attention: The effect of visual onsets and offsets. Perception and Psychophysics, 49, 83–90.
Theeuwes, J., Kramer, A.F., Hahn, S. & Irwin, D.E. (1998). Our eyes do not always go where we want them to go: Capture of eyes by new objects. Psychological Science, 9, 379–385.
Watson, D.G. & Humphreys, G.W. (1995). Attention capture by contour onsets and offsets: No special role for onsets. Perception and Psychophysics, 57, 583–597.
Watson, D.G. & Humphreys, G.W. (1997). Visual marking: Prioritizing selection for new objects by top-down attentional inhibition of old objects. Psychological Review, 104, 90–122.
Watson, D.G., Humphreys, G.W. & Olivers, C.N.L. (2003). Visual marking: Using time in visual selection. Trends in Cognitive Sciences, 7(4), 180–186.
Yantis, S. & Jonides, J. (1984). Abrupt visual onsets and selective attention: Evidence from visual search. Journal of Experimental Psychology: Human Perception and Performance, 10, 601–621.
Yantis, S. & Jonides, J. (1990). Abrupt visual onsets and selective attention: Voluntary versus automatic allocation. Journal of Experimental Psychology: Human Perception and Performance, 16, 121–134.