The present results clearly demonstrate that a fundamental aspect of visual perception, the parsing of temporal patterns, has limited capacity and therefore critically depends on the availability of attention. The level of perceptual load in a spatial visual search determined the accuracy and the sensitivity of flicker perception at fixation.
These findings also generalize load theory (Lavie,
1995,
2005) to the temporal domain. Previous studies have only shown effects of perceptual load on the conscious perception of spatial patterns. Recent examples include demonstrations that the level of perceptual load in similar search tasks to that used here determines the rates of “change blindness” (Beck & Lavie,
2007; Lavie,
2006), “inattentional blindness” for an unexpected shape (Cartwright-Finch & Lavie,
2006), and detection sensitivity for expected shapes (Carmel, Rees, & Lavie,
2006; Macdonald & Lavie,
2007). Here we extend the evidence for load theory to show that the level of perceptual load in a visual search task determines the conscious detection of a temporal pattern.
Although these findings are consistent with load theory, they appear to be at odds with a recent finding (Yeshurun,
2004; Yeshurun & Levy,
2003) that cueing attention to the stimulus location impairs, rather than facilitates, temporal resolution (assessed by sensitivity to the rapid disappearance and reappearance of a stimulus, a method known as double pulse resolution, DPR). Yeshurun and colleagues obtained their results using spatial cueing, whereas we used a visual search paradigm to vary load on attention. However, as both the spatial cuing paradigm and our manipulation of load in visual search serve to vary the allocation of attention, the contrast in results may at first sight appear surprising, whereas directing attention with a spatial cue seems to impair temporal resolution; our findings clearly demonstrate that such resolution benefits from the allocation of attention. Our results are also hard to reconcile with the theoretical account proposed by Yeshurun and Levy (
2003). They suggested that spatial cueing facilitates the activity of parvocellular neurons in retinotopic regions corresponding to the attended location, which in turn leads to inhibition of magnocellular neurons at the same location. This impairs temporal resolution at attended locations due to the longer response durations of parvocellular neurons (Merigan & Maunsell,
1993; Schiller & Logothetis,
1990). Clearly, our findings cannot be accounted for by this mechanism. If the greater availability of attention under low perceptual load was mediated by an increase in parvocellular activity, this account would predict impaired (rather than improved) temporal resolution under low (compared to high) load. It is therefore likely that rather than the availability of attention per se being the element common to both our paradigm and that of Yeshurun and colleagues, the different results are instead due to the different way in which availability of attention was varied between the two paradigms. Specifically, the spatial cueing task employed by Yeshurun (
2004) and Yeshurun and Levy (
2003) invoked transient attention, known to be reflexively drawn to the cued location for a limited duration (less than 250 ms; Carrasco & McElree,
2001). Such reflexive shifts of attention toward a peripheral cue may interfere with DPR because of the involvement of transients in such cueing and in the detection of the pulse onset and offset. The level of perceptual load in the task used here affects the allocation of attention to the fixated flickering LED without involving such transient reflexive shifts. In this case, the lower demands on attentional resources under low load clearly benefit temporal resolution.
This possibility is further supported by another recent study (Poggel, Teutwein, Calmanti, & Strasburger,
2006), which used the DPR method to show that temporal resolution at fixation is improved when the spatial extent of the attentional focus is reduced. Poggel et al. (
2006) asked participants to detect a double pulse in one of nine stimuli (one at fixation and eight arranged in a peripheral circle). The eccentricity (though not the size) of the peripheral stimuli was varied. Temporal resolution for the fixated stimulus was better when the peripheral stimuli were at a smaller eccentricity. Here the allocation of spatial attention varied without involving any transient reflexive shifts, and temporal resolution benefited from greater availability of attention.
The results of Poggel et al. (
2006) are consistent with load theory (Lavie,
1995,
2005): When attending to both fixated and peripheral locations, increasing the eccentricity of peripheral stimuli leads to increased demands on divided spatial attention. This can be construed as an increase in perceptual load, since both visual acuity and DPR sensitivity decline as eccentricity increases. Importantly, however, the present study demonstrates the effect of perceptual load on temporal resolution independently of the extent of spatial attention, as the eccentricity of stimuli in the letter search was identical for both load conditions. Perceptual load was therefore dissociated from spatial attention by keeping the spatial extent of divided attention constant. Future research may attempt to further eliminate the involvement of spatial attention in the effects of perceptual load by varying load for stimuli that spatially overlap with the flickering stimulus.
The effect of perceptual load in a spatial latter search on flicker detection suggests that temporal pattern perception shares capacity limits with spatial information processing. Though the luminance contrast of the flickering light was kept constant throughout the study, the high contrast used was clearly supra-threshold. This makes it is unlikely that the effects of perceptual load on the perception of near-threshold temporal frequency was mediated by any effect of perceptual load on contrast perception for the LED.
What is the neural resource shared by temporal and spatial processes? Although our behavioral data cannot directly implicate any neural substrates, previous research facilitates speculation on the processes involved. Imaging studies have associated the level of neural activity in posterior parietal cortex with the effects of perceptual load (e.g., Schwartz et al.,
2005; Wojciulik & Kanwisher,
1999) and neuropsychological work suggests that neglect patients with parietal lesions have reduced perceptual capacity overall (Lavie & Robertson,
2001), not only in spatial tasks (Cusack, Carlyon, & Robertson,
2000). Indeed, our recent neuroimaging work (Carmel, Lavie, & Rees,
2006) showed that flicker detection is associated with activity in similar parietal regions to those implicated in neglect and the effects of perceptual load. These regions included the intraparietal sulcus and inferior parietal lobule. Though anatomical overlap at the gross scale of fMRI does not necessarily imply shared neural circuitry or function, it raises the possibility that the same attentional resources, mediated by activity in the above areas, may be recruited for both spatial search tasks and flicker detection. The specific neural mechanism underlying such recruitment awaits characterization. However, it is likely to mediate the allocation of limited-capacity attention to sensory information received from early visual cortex, enabling conscious access to such information (for reviews of the involvement of overlapping regions of frontal and parietal cortex in both awareness and attention, see Naghavi & Nyberg,
2005; Rees, Kreiman, & Koch,
2002).
Finally, it is important to note that although we have shown effects of perceptual load on conscious perception of flicker (as expressed by the explicit reports of our participants), we by no means suggest that these effects are confined to conscious representations of flicker. It remains possible that perceptual load can affect earlier, preconscious processing of flicker. Indeed, the effects seen at the conscious level may be due to the effects of perceptual load on such earlier levels of flicker processing; recent research (Bahrami, Carmel, Walsh, Rees, & Lavie,
2007; Bahrami, Lavie, & Rees,
2007) shows that perceptual load affects both unconscious perception of task-irrelevant stimuli (e.g., orientations, or images of tools) and the activity they evoke in V1. Whether perceptual load in a spatial search can affect early unconscious processing of flicker remains an interesting topic for future research.