Limits...
An analysis of visual masking, with a defense of 'Stopped Processing'.

Reeves A - Adv Cogn Psychol (2008)

Bottom Line: This paper provides evidence for stopped processing and some applications of this to object recognition and letter detection.The paper also discusses the notion of an 'active filter' which may help to account for Type-A masking but at best can only account for Type-B masking in part.I conclude that masking, while illuminating various areas of vision science, is under-utilized, perhaps because the theoretical justification for such masking is still uncertain, and perhaps because of the care needed to establish that the mask does indeed 'stop' processing.

View Article: PubMed Central - PubMed

Affiliation: Dept. of Psychology, Northeastern University, Boston, USA.

ABSTRACT
The use of a backward mask (a patterned mask which follows the target in time) to 'stop the processing' of the target illustrates an important application of masking - the study of the 'microgenesis' of visual perception, that is, visual processing over about the first one-fifth of a second. This paper provides evidence for stopped processing and some applications of this to object recognition and letter detection. The paper also discusses the notion of an 'active filter' which may help to account for Type-A masking but at best can only account for Type-B masking in part. I conclude that masking, while illuminating various areas of vision science, is under-utilized, perhaps because the theoretical justification for such masking is still uncertain, and perhaps because of the care needed to establish that the mask does indeed 'stop' processing.

No MeSH data available.


The mean number of disks reported as a function of the number presented (Np)						in NM (black rectangles) and in BM at two SOAs, 30 ms (BM1: closed circles)						and 50 ms (BM2: open circles). Values of B, the capacity limit in masking,						were chosen to best-fit the data to the interruption predictions, separately						in BM1 and BM2. Data follow the predictions of integration and interruption						fairly closely, although averaging over individual participants smoothed the						BM curves a little.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2864976&req=5

Figure 2: The mean number of disks reported as a function of the number presented (Np) in NM (black rectangles) and in BM at two SOAs, 30 ms (BM1: closed circles) and 50 ms (BM2: open circles). Values of B, the capacity limit in masking, were chosen to best-fit the data to the interruption predictions, separately in BM1 and BM2. Data follow the predictions of integration and interruption fairly closely, although averaging over individual participants smoothed the BM curves a little.

Mentions: In our attempt to distinguish stopped processing from integration (Liss & Reeves, 1983), participants reported the number of black disks, from 0 to 10, presented at random locations within an 8-by-8 grid on a white screen. In NM, the disks were presented near-threshold by flashing them for just 2 or 3 ms to reduce their effective contrast. In BM the disks were presented at full contrast for 20 ms, but followed after a variable period by a 200 ms duration, patterned mask (Fig. 1, top). The mask was an 8 x 8 array of disks just slightly bigger than the target disks, located in the same positions as the target disks so that the masking was of the form ‘backward masking by pattern’ rather than ‘metacontrast’ in nature. The graph in Figure 1 presents the theoretical predictions of integration and interruption. Integration predicts that for low contrasts, the target integrates with the white field so the visibility of the disks is reduced. If the disks are far enough apart to eliminate lateral interactions, their chances of being seen are independent of one another. Thus the number of reported disks will be proportional to the number presented, at least until the number of items begins to exceed the short-term memory span. In contrast, the interruption theory predicts that the mask ‘stops processing’ when presented. In the (ideal) example plotted, the mask stops processing after 3 target disks have been encoded, so that performance is perfect for 0, 1, 2, and 3 disks, but no more than 3 disks are ever reported. The data for 6 subjects each clearly followed the integration prediction for NM, which accounted for 92% of the variance, and the interruption prediction for BM, which accounted for 90% of the variance (Fig. 2). In contrast, integration accounted for only 59% of the variance of the BM data, so the results clearly support the interruption model for BM over integration. This pattern of results was repeated for both strict and lax criteria for reporting a disk.


An analysis of visual masking, with a defense of 'Stopped Processing'.

Reeves A - Adv Cogn Psychol (2008)

The mean number of disks reported as a function of the number presented (Np)						in NM (black rectangles) and in BM at two SOAs, 30 ms (BM1: closed circles)						and 50 ms (BM2: open circles). Values of B, the capacity limit in masking,						were chosen to best-fit the data to the interruption predictions, separately						in BM1 and BM2. Data follow the predictions of integration and interruption						fairly closely, although averaging over individual participants smoothed the						BM curves a little.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2864976&req=5

Figure 2: The mean number of disks reported as a function of the number presented (Np) in NM (black rectangles) and in BM at two SOAs, 30 ms (BM1: closed circles) and 50 ms (BM2: open circles). Values of B, the capacity limit in masking, were chosen to best-fit the data to the interruption predictions, separately in BM1 and BM2. Data follow the predictions of integration and interruption fairly closely, although averaging over individual participants smoothed the BM curves a little.
Mentions: In our attempt to distinguish stopped processing from integration (Liss & Reeves, 1983), participants reported the number of black disks, from 0 to 10, presented at random locations within an 8-by-8 grid on a white screen. In NM, the disks were presented near-threshold by flashing them for just 2 or 3 ms to reduce their effective contrast. In BM the disks were presented at full contrast for 20 ms, but followed after a variable period by a 200 ms duration, patterned mask (Fig. 1, top). The mask was an 8 x 8 array of disks just slightly bigger than the target disks, located in the same positions as the target disks so that the masking was of the form ‘backward masking by pattern’ rather than ‘metacontrast’ in nature. The graph in Figure 1 presents the theoretical predictions of integration and interruption. Integration predicts that for low contrasts, the target integrates with the white field so the visibility of the disks is reduced. If the disks are far enough apart to eliminate lateral interactions, their chances of being seen are independent of one another. Thus the number of reported disks will be proportional to the number presented, at least until the number of items begins to exceed the short-term memory span. In contrast, the interruption theory predicts that the mask ‘stops processing’ when presented. In the (ideal) example plotted, the mask stops processing after 3 target disks have been encoded, so that performance is perfect for 0, 1, 2, and 3 disks, but no more than 3 disks are ever reported. The data for 6 subjects each clearly followed the integration prediction for NM, which accounted for 92% of the variance, and the interruption prediction for BM, which accounted for 90% of the variance (Fig. 2). In contrast, integration accounted for only 59% of the variance of the BM data, so the results clearly support the interruption model for BM over integration. This pattern of results was repeated for both strict and lax criteria for reporting a disk.

Bottom Line: This paper provides evidence for stopped processing and some applications of this to object recognition and letter detection.The paper also discusses the notion of an 'active filter' which may help to account for Type-A masking but at best can only account for Type-B masking in part.I conclude that masking, while illuminating various areas of vision science, is under-utilized, perhaps because the theoretical justification for such masking is still uncertain, and perhaps because of the care needed to establish that the mask does indeed 'stop' processing.

View Article: PubMed Central - PubMed

Affiliation: Dept. of Psychology, Northeastern University, Boston, USA.

ABSTRACT
The use of a backward mask (a patterned mask which follows the target in time) to 'stop the processing' of the target illustrates an important application of masking - the study of the 'microgenesis' of visual perception, that is, visual processing over about the first one-fifth of a second. This paper provides evidence for stopped processing and some applications of this to object recognition and letter detection. The paper also discusses the notion of an 'active filter' which may help to account for Type-A masking but at best can only account for Type-B masking in part. I conclude that masking, while illuminating various areas of vision science, is under-utilized, perhaps because the theoretical justification for such masking is still uncertain, and perhaps because of the care needed to establish that the mask does indeed 'stop' processing.

No MeSH data available.