The role of eye fixations in concentration and amplification effects during multiple object tracking

Abstract

When tracking spatially extended objects in a multiple object tracking task, attention is preferentially directed to the centres of those objects (attentional concentration), and this effect becomes more pronounced as object length increases (attentional amplification). However, it is unclear whether these effects depend on differences in attentional allocation or differences in eye fixations. We addressed this question by measuring eye fixations in a dual-task paradigm that required participants to track spatially extended objects, while simultaneously detecting probes that appeared at the centres or near the endpoints of objects. Consistent with previous research, we observed concentration and amplification effects: Probes at the centres of objects were detected more readily than those near their endpoints, and this difference increased with object length. Critically, attentional concentration was observed when probes were equated for distance from fixation during free viewing, and concentration and amplification were observed without eye movements during strict fixation. We conclude that these effects reflect the prioritization of covert attention to particular spatial regions within extended objects, and we discuss the role of eye fixations during multiple object tracking.

Acknowledgements

For helpful conversation and/or comments on earlier drafts, we thank Jason Reiss, George Alvarez, Jon Flombaum, Todd Horowitz, and two anonymous reviewers. This work was supported in part by NIH subaward 8508-53692 from Johns Hopkins University to JEH. BJS was supported by NSF No. BCS-0132444. Some of these data were presented at the 47th annual meeting of the Psychonomic Society, Houston, Texas, November 2006.

Notes

¹Note that this possibility does not imply that participants strategically fixate the objects’ centres while tracking spatially extended objects. Rather, our contention is that object centres may be preferentially fixated simply as a result of the normal operation of saccadic eye movements. In fact, Alvarez and Scholl (2005) ruled out potential contributions of higher level strategies in producing concentration and amplification effects by showing robust concentration and amplification under conditions in which such strategies would have favoured the detection of end probes. In one case, the participants’ task was to track the endpoints of spatially extended objects; in another case, end probes appeared with a higher probability than centre probes. If attention and/or fixations had been strategically directed to objects’ ends in these cases then one would have expected to observe a detection advantage for end probes relative to centre probes—but in fact the concentration and amplification effects were observed in these conditions.

²One puzzling aspect of these results is that concentration effect (and amplification effect, as reported later) occurred for the distractors as well as the targets, even though observers had no obvious incentive to attend to distractors. This was also true in the studies of Alvarez and Scholl (2005), and it is reminiscent of the finding that probes presented on MOT targets were easier to detect than probes occurring on distractors—which, in turn, were harder to detect than probes presented in empty space (Pylyshyn, 2006; see also Flombaum & Scholl, 2008; Flombaum, Scholl, & Pylyshyn, 2008). Observers are apparently “paying attention” in some way to distractors, if only to suppress them (Pylyshyn, 2006) or to mark some of them as distractors in order to improve guessing performance (Hulleman, 2005). The fact that both targets and distractors show the same nonuniform distribution of attention over the shape of the object suggests that the amplification and concentration aspects of object-based attention may be separable from other processes responsible for determining task relevance.

³Observers in Alvarez and Scholl (2005) were not required to maintain fixation, but in the Supplementary Appendix to that paper, these authors nevertheless showed that the concentration and amplification effects persisted when probes at objects’ centres versus ends were equated for their distance from the centre of the display.

⁴In order to ensure that the observed concentration effect was not produced by inadvertent differences in probe eccentricity produced by random probe presentations, we performed a binned analysis similar to that of Experiment 1. A 2 (Probed object: Target vs. distractor)×2 (Probe position)×4 (Probe-fixation distance) repeated-measures ANOVA revealed a significant interaction between probe position and probe-fixation distance, F(3, 42) = 4.44, MSE = 0.074, p<.05. As was the case for target probes in Experiment 1, significant concentration was not observed for probes presented very near to fixation, t(14) = 0.81, Bonferroni-corrected p>.99, but significant concentration was observed for each of the other three probe-fixation distances: Near, t(14) = 3.56, p<.05; far, t(14) = 7.51, p<.001; very far, t(14) = 5.43, p<.001; Bonferroni corrected ps.

⁵In order to ensure that the observed amplification effect was not produced by inadvertent differences in probe eccentricity produced by random probe presentations, we performed a binned analysis similar to that of Experiment 1. A 2 (Probed object)×2 (Probe position)×2 (Object length)×2 (Probe-fixation distance) repeated-measures ANOVA revealed a significant interaction between probe position and probe-fixation distance, F(3, 42) = 4.44, MSE = 0.074, p<.05, indicating that the size of the concentration effect increased with object length. In other words, reliable amplification was observed.