Patrick T. Goodbourn1, Paolo Martini2, Michael Barnett-Cowan3,4, Irinia M. Harris1, Evan J. Livesey1, Alex O. Holcombe1
1School of Psychology, University of Sydney, Australia 2Department of Psychology, University of Warwick, United Kingdom 3Department of Psychology, University of Western Ontario, Canada 4Department of Kinesiology, University of Waterloo, Canada
Two attentional episodes cannot occur very close in time. This is the traditional theory of the attentional blink, and it correctly predicts that the second of two successive attentional episodes often fails. But even when an episode succeeds, it may occur at an inappropriate time. Based on an analysis of response errors, Vul, Nieuwenstein and Kanwisher (2008) concluded that selection associated with a second target (T2) was temporally advanced for short lags and delayed for longer lags, and was less temporally precise during the blink period. However, their parametric estimates of attentional episode characteristics can be biased by instances in which the item reported for T2 was selected during an episode directed at the first target (T1). Such instances are evident in response error distributions, and could explain the phenomenon of lag-1 sparing. We reanalysed data from six studies, using mixture modelling to assess the characteristics of attentional episodes. At each lag, we compared two models: the first assumed that both target reports (T1 and T2) were drawn from a single attentional episode directed at T1; the second included an additional episode directed at T2. The results suggest that a second episode occurs only if lag exceeds 100–250 ms, with the probability of initiating an episode returning to baseline for lags beyond about 500 ms. When a second episode does occur, the magnitude of its delay decreases as lag increases; but its temporal precision is invariant with lag, and is indistinguishable from a T1 baseline. This confirms that second attentional episodes are suppressed and delayed, but suggests that they are not temporally advanced for short lags, and that their temporal precision is not affected by earlier episodes. It also suggests that at least two items are sometimes retrieved from the first attentional episode, explaining lag-1 sparing.
Prism adaptation ameliorates pseudoneglect by enhancing target processing in right hemispace
Elizabeth Nguyen, Patrick T. Goodbourn, Alex O. Holcombe
Line bisection and related tasks reveal a leftward attentional bias in healthy individuals. This bias is reduced by adaptation to visual experience of the scene shifted to the left. One theory is that a balance of activation between the two cerebral hemispheres—usually favoring the right hemisphere—is shifted after prism adaptation. Hence, an increase in processing on one side should be accompanied by a decrease on the other. Standard methods, such as line bisection, yield only relative measures of left–right performance; thus to test the theory, we instead derived independent, absolute measures of performance for left and right hemispace. Because it has been proposed that the two hemispheres are differentially involved in temporal processing, we also assessed whether adaptation affected temporal parameters of attentional selection. Goodbourn & Holcombe (VSS 2013) found a leftward bias, consistent with pseudoneglect, when two rapid serial visual presentation (RSVP) streams of letters were presented concurrently to the left and right of fixation. Target letters were cued simultaneously in both streams. While participants frequently reported simultaneous letters, efficacy of target selection was substantially higher for the left stream than for the right. In the present study, twelve observers completed the dual-RSVP task before and after adaptation to three types of prisms: left-shifting, right-shifting, and control. Consistent with previous pseudoneglect studies, no performance changes were found after right-shifting or control prism adaptation. In contrast, adaptation to left-shifting prisms reduced the leftward bias, selectively boosting right-stream performance: relevant letters were reported 9% more often after adaptation ( = 0.03). We found no accompanying decrement in left-stream performance. Temporal parameters (latency and precision) were unchanged for both sides. These findings suggest prism adaptation ameliorates biases not by rebalancing hemispheric activation, but rather by selectively enhancing processing of stimuli in the non-dominant right hemifield.
Dividing attention reduces both speed and temporal frequency limits on object tracking
Alex Holcombe, Wei-Ying Chen
Low-level motion processing is limited by temporal frequency, not speed: for a broad range of spatial frequencies, the fastest speeds at which motion is perceived correspond to a single temporal frequency, meaning that the threshold speed is proportional to the inverse of spatial frequency (Burr & Ross, 1982). Like low-level motion, attentional tracking also has a temporal frequency limit, but one that is much lower (7 Hz) and declines progressively with more targets to track. Also unlike low-level motion, attentional tracking additionally has a speed limit (Holcombe & Chen, 2013 and Verstraten et al., 2000), supporting a qualitative difference between tracking processes and low-level motion. Here, one tracking target revolved about the fixation point, along with a distractor or distractors sharing the same circular trajectory. When the number of distractors is higher than about four, and speed is increased, performance falls to threshold when a particular temporal frequency is reached. But for fewer distractors, the limiting factor instead appears to be speed. Like Verstraten et al. (2000) and Holcombe & Chen (2013), for one target we observe a limit of about 2 revolutions per second (rps). Here we varied the number of targets by presenting three concurrent, concentric displays. A second target reduced the speed limit by 0.4 rps, and a third target by a further 0.5 rps. Thus, dividing attention reduces tracking’s speed limit substantially, just as it does its temporal frequency limit. Unlike the temporal frequency limit, the speed limit was not robust to changes in the number of distractors sharing the trajectory. When two rather than one distractor shared each target’s trajectory, the speed limits fell by 0.5 rps. Theories of tracking must be modified if they are to explain how dividing attention reduces both limits, and additionally explain the decrease in speed limits caused by an additional distractor.
The attentional blink in right parietal patients: Analysis of temporal selection parameters
Lorella Battelli, Sara Agosta, Paolo Martini, Alex O. Holcombe, Patrick T. Goodbourn
Patients affected by right parietal lesions have impairments in visual timing tasks (Battelli et al. 2008) and may also show an extended attentional blink (AB, Husain et al. 1997), an impairment in detecting the second of two targets appearing in close temporal succession. This overall disturbance of temporal attention may have distinct components that are differentially altered in patients. We used a rapid serial visual presentation (RSVP) display containing a stream of 26 letters presented at 8.6 items/sec. Two letters were targets cued by an annulus, and the number of items (the lag) between the first target (T1) and second target (T2) was varied. Participants were asked to report both targets. For each participant we measured accuracy for T1, and for T2 contingent on correct report of T1. By analyzing the serial position of non-target items reported as targets, we also derived estimates for three distinct components of attentional selection: (i) efficacy, comparable to the probability of reporting an item in a time window around the target; (ii) latency; and (iii) temporal precision. We tested three right parietal patients (FC, RR and PP). All showed an extended AB—in terms of both contingent T2 accuracy (i.e. correct) and T2 efficacy ( correct)—replicating previous findings. FC and PP showed low T1 accuracy but normal T1 efficacy; in contrast, RR showed both low accuracy and low efficacy. For PP and RR, T1 and T2 selections were systematically delayed and less temporally precise relative to controls, whereas latency and precision were relatively spared for FC. Interestingly, patient PP showed lag-1 sparing, which was not evident for FC nor for RR. These results confirm that parietal patients show an extended blink, but point to substantial individual differences in component attentional processes. We discuss how these differences may relate to lesion site.