Modulating Irrelevant Motion Perception by Varying Attentional Load in an Unrelated Task

See allHide authors and affiliations

Science  28 Nov 1997:
Vol. 278, Issue 5343, pp. 1616-1619
DOI: 10.1126/science.278.5343.1616


Lavie's theory of attention proposes that the processing load in a relevant task determines the extent to which irrelevant distractors are processed. This theory was tested by asking participants in a study to perform linguistic tasks of low or high load while ignoring irrelevant visual motion in the periphery of the display. Although task and distractor were unrelated, both functional imaging of motion-related activity in cortical area V5 and psychophysical measures of the motion aftereffect showed reduced motion processing during high load in the linguistic task. These findings fulfill the prediction that perception of irrelevant distractors depends on the relevant processing load.

To what extent does perception depend on attention? This issue has been a central question in attention theory over the past 40 years, yet it remains unresolved. Two contrasting positions have emerged. Some studies have suggested the importance of attention for perception, showing that unattended stimuli apparently receive very little processing (1). Other studies, however, have implied that unattended stimuli can be perceived and have some effect on behavior as measured by indirect methods (for example, reaction times and evoked potentials) (2). Here, we combine functional imaging and psychophysics to test a theory that resolves the long-standing controversy between these two established positions.

Lavie's theory (3) proposes that capacity for perception is limited but that, within those limits, perception proceeds automatically. Thus, although we may not be able to perceive everything, we are unable to stop perceiving whatever we can. The extent to which a target task exhausts available capacity thus determines the extent to which irrelevant distracting stimuli will be processed. If the processing load of the target task exhausts available capacity, irrelevant stimuli will not be perceived. However, if the target-processing load is low, attention will inevitably spill over to the processing of irrelevant distractors (4). The extent to which irrelevant stimuli are excluded from perception does not thus depend simply on participants' intentions to ignore them. Irrelevant stimuli are excluded from perception only when the processing load of a task engages full attention under conditions of high load (5).

We sought to investigate this theory by studying the perception of irrelevant visual motion during performance of a task requiring linguistic judgments on single words. Although both of these tasks require visual input, they are thought to rely on completely different psychological processes. Yet, if both depend on a common source of attention, as the load theory suggests, then they should be strongly interdependent. Specifically, we predicted that participants would fail to ignore irrelevant visual motion as long as processing load in the linguistic task was low but that higher load in the linguistic task would prevent perception of the irrelevant motion. Despite participants' wishes to ignore the motion distractors in all conditions, they would only succeed in doing so under conditions of high load in another task that exhausted their attentional capacity.

Visual motion was used as a distracting stimulus as it is known to activate a distinct area of the brain, V5, whose location has been reliably identified in previous functional-imaging studies (6, 7). Activation of V5 by a moving stimulus should therefore allow determination of whether processing of irrelevant visual motion has occurred. Previous functional-imaging, psychophysical, and single-cell electrophysiological studies have all suggested that motion perception may depend to some extent on attention (8-12). However, none of these studies have provided a critical test for our claim that the crucial factor determining when participants can ignore motion distractors is the attentional load in an unrelated task. Every previous study has compared explicit attention to motion with explicit ignoring of motion. Any difference in motion-related brain activity between these two conditions can be attributed to an enhancement of perception with deliberate attention to the moving stimulus, rather than successful ignoring of motion in the unattended condition. These previous studies therefore cannot provide any clear answer to the principal issue in attention theory: whether irrelevant distractors can be excluded from perception. Here, we test the load theory by characterizing changes in selective processing of a motion stimulus that is always irrelevant, while varying the attentional load of an unrelated target task.

For the irrelevant motion stimulus, we used an optic flow field with a full field of dots moving radially toward the screen edge. This type of motion may be particularly difficult to ignore because of its biological relevance (13). Irrelevant motion processing was characterized by measurement of brain activity in motion-related areas during changes in the demands of an imposed (and unrelated) linguistic task. Participants viewed a display with two different components. In the periphery of the display were scattered white dots. We assessed motion perception by comparing conditions under which these dots moved with conditions under which the dots were static (14). Participants were asked to ignore the white dots throughout and were told that the dots were always irrelevant to the experiment and might produce unpleasant motion aftereffects if they were not ignored. In the center of the display, single words were presented successively in a blank ellipse that separated them from the dots (15). Participants were asked to focus on these words and, during low-load conditions, press a key whenever a word was printed in uppercase letters. Under high-load conditions, they saw the same letter strings but were now asked to press the button whenever they saw a bisyllabic word. Each participant performed both high- and low-load tasks, with and without irrelevant visual motion, while undergoing functional magnetic resonance imaging (fMRI) (16). We confirmed that processing load was manipulated appropriately by recording participants' responses in the linguistic task (17).

The critical test of our hypothesis is whether evoked activity related to irrelevant visual motion (compared with no motion) is smaller under conditions of high processing load (compared with low load). This pattern of modulation is represented by the interaction term in the factorial design of the experiment. We therefore constructed the statistical parametric map that reflects this interaction between processing load and visual motion, using statistical parametric mapping (SPM) (18, 19). Data from all six participants were analyzed as a group to identify areas activated in common across all participants. This analysis identified several areas (Table 1) in which the effect of visual motion (compared with no motion) was greater under conditions of low load (compared with high load). Our discussion will be limited to those areas at or before V5 in the pathway for the processing of visual motion (6, 7,20).

Table 1

Areas where evoked activity during visual motion (compared with the no-motion conditions) was significantly greater under conditions of low load (compared with high load) (19,20). Only areas that are also active during the comparison of visual motion (irrespective of load) and rest (fixation) are shown (20), to ensure that only areas concerned with the processing of visual motion are considered. Only areas that reachP < 0.05 after correction for multiple comparisons are reported, except in V5, where a threshold of P < 0.001 uncorrected was used (because of our previous anatomical hypothesis for this area).

View this table:

Robust bilateral modulation of V5 complex activity, related to visual motion by load in the target task, was identified (Fig.1) (21). This interaction of motion and load is in accord with our experimental hypothesis. Under conditions of low load, the moving dots produced strong activation compared with the static dots (and baseline fixation), suggesting motion perception. However, under conditions of high load, there was no increase in activity associated with the moving dots. Thus, under conditions of high load, we infer that distracting visual motion was not processed, whereas under low-load conditions it was processed. In other words, despite our instructions to participants to always ignore the dots, their perception of irrelevant visual motion was in fact determined by the experimental manipulation of load in an unrelated task and not by their intentions alone.

Figure 1

(A and B) Lateral views of the right and left hemispheres of a T1-weighted volume–rendered anatomical image that conforms to the stereotactic space of Talairach and Tournoux. Superimposed in red are the areas from Table 1 where brain activity in the group of participants showed the predicted interaction between the effects of visual motion and linguistic-processing load. The locations of the right and left V5 complex activity described in the text and Table 1 are indicated by the arrows. (C) A sagittal slice through the same canonical anatomical image, on which is superimposed the location of activity in the SC (arrow) that is due to the interaction of visual motion and linguistic-processing load. (D and E) Mean activity over all participants and replications of each experimental condition taken from the left V5 complex area described in Table 1. Activity during baseline periods (dark gray shading) is shown alternating with that during experimental conditions (light gray shading). The order in which the conditions are displayed is illustrative and does not correspond to that used in the experiment (because order of conditions was counterbalanced across participants). The statistical comparisons reported in Table 1 and the text refer to the comparison of the experimental conditions (light gray). The scale bar represents a value of 0.1% BOLD signal change.

We tested this result further in a second experiment that was adapted from Chaudhuri's psychophysical procedure (8) to make it suitable for testing the load theory. Prolonged exposure to visual motion, followed by viewing of a static stimulus, produces an illusory perception of opposing motion in the static display that fades over time. This motion aftereffect is contingent on V5 activity (22) and has been shown to be sensitive to attention (8, 9). The duration of this aftereffect can therefore serve as a behavioral probe for the extent of irrelevant visual motion processing in our task, allowing a convergent test of our load hypothesis. The pattern of evoked activity observed in V5 complex suggests that the motion aftereffect should be substantially reduced under high-load conditions of the linguistic task. Four participants viewed displays identical to those used in the functional-imaging experiment while performing either the high- or low-load task. The surround was always moving and was followed by a static full field of dots that produced a vivid motion aftereffect, where the static dots appeared to radially contract. This motion aftereffect was significantly shorter under conditions of high load than under conditions of low load for all participants (23). This result is consistent with both our theoretical predictions and our observation in the fMRI study that motion-specific responses were suppressed with high load in the linguistic task.

In the functional-imaging data, a number of other areas, predominantly visual, showed a significant modulation of motion-related activity by processing load in addition to V5. Our results suggest that the effects of processing load become manifest at multiple levels of the sensorimotor network, including very early visual areas. Differential activation was seen at the V1-V2 border on the right and at a lower significance on the left. Reciprocal connections between V1-V2 and V5 have been demonstrated by reversible cooling of V5 and neuroanatomical studies in monkey (24). Changes in neural responses in early visual cortex that are due to attention have previously been shown in monkey (25, 26); our results show that early visual areas in humans are also sensitive to the effects of attention. Modulation of motion-related activity by attentional load was also seen in the superior colliculus (SC) (Fig.1). This pattern of evoked responses is compatible with an ablation study in monkey that demonstrated that the impairment in visual discrimination after a lesion in the SC is manifest only when the unaffected part of the visual field contained a competing item (27). Our data suggest that the SC is sensitive to attentional load (rather than just low-level visual competition between stimuli), because we modulated motion-related activity in the SC in our study by varying the processing requirements in a target task without adding more visual stimuli. Thus, there is greater competition for attention with either an increased number of items or with more processing for the same items (4, 26). Moreover, the SC has direct reciprocal anatomical connections with V5, and both the SC and the striate cortex contribute to visual function and motion-specific neural response properties in V5 (28). It is interesting, therefore, that we observe modulation of motion processing by load in an unrelated cognitive task in both the geniculostriate (V1-V2) and retinotectal (SC) pathways by which motion-related signals can reach V5. Although the SC has been implicated in oculomotor control, eye movements do not provide a plausible explanation for our findings (29).

In conclusion, our results demonstrate the use of functional imaging to test a cognitive theory of attention. Specifically, we have proposed a resolution to the long-standing issue of whether perception of irrelevant stimuli depends on attention. Our results show that participants' intentions to avoid irrelevant distractors are not always sufficient for ignoring them. As long as the target task imposes only a low load on attention, irrelevant stimuli such as motion will still be perceived. However, this irrelevant perception is strongly reduced if the load of the unrelated task is increased. Selective perception is therefore possible only under conditions of high load. Under such conditions, even the perception of biologically significant stimuli such as optic flow can be reduced by a demanding but entirely unrelated task.

  • * To whom correspondence should be addressed. E-mail: g.rees{at}


View Abstract

Navigate This Article