Spatial interference with relative localisation 20

10 - -10 -20 3 4 5 6 7 8 9 1 0

flanking distance (min arc)

Figure 3.29: Mean shifts are shown for edge targets, with a single Gaussian bar flank. Data is shown for two subjects (IRP; open circles and GLC; filled circles). The flanking bar had a contrast 50% that of the edge and had a positive polarity- of

contrast.

100 q

X i

a.

3 4 5 6 7 8 9 1 0

flanking distance (min arc)

Figure 3.30: Localisation thresholds are shown for edge targets, with a single Gaussian bar flank with positive contrast polarity (50% the contrast of the edge). Data is shown for two subjects (IRP; open circles, and GLC; filled circles). Again, there is no clear dependency of precision of localisation on the separation of the target edges and the flanking bar. The edges and bar had space constants of 1.0 min arc.

Chapter 3: Spatial interference with relative localisation

Mean shifts values and thresholds obtained are displayed in figures 3.27-3.30 as a function of the flanking distance between the target cumulative Gaussian edge and flanking Gaussian bar. The results are unlike those presented for line-line interactions (Badcock & Westheimer, 1985a,b) or in the previous experiments, for either the bar- bar or bar-edge interactions. It is evident that the effects of polarity of contrast of the flanking feature are reversed if the feature to be localised is an edge. Badcock and Westheimer demonstrated only repulsion of lines if the flank and target are opposite in contrast, but in the present experiment, attraction is seen between a positive contrast blurred edge and a negative contrast blurred bar. A positive contrast blurred edge is repelled by a positive contrast blurred bar.

This indicates that the interaction between bars and edges is asymmetric: in the next experiment, it was demonstrated that a positive contrast target bar was attracted towards a positive contrast flanking edge. The present experiment demonstrates that this must be accompanied by a simultaneous shift in the perceived location of the edge away from the target bar in these circumstances.

In comparison with the data from a bar flanked by an edge, an edge which is to be localised is distorted in position more by an adjacent bar, than a bar to be localised which is flanked by an edge. The shifts in perceived location were reliably larger when the feature to be localised was an edge, rather than a bar; partly, this may reflect the higher thresholds for edge localisation (Watt & Morgan, 1983a), although subject's phenomenal reports suggested that the edge localisation was genuinely impaired more by a flanking feature than bar localisation.

88

1

Chapter 3: Spatial interference with relative localisation

3.4.4: Experiment 4: Vernier acuity for Gaussian bars with ed^e flanks

10

-,

-10

3 4 5 6 7 8 9 1 0

flanking distance (mln arc)

Figure 3.31: Location biases are shown for two observers for positive contrast Gaussian bar targets flanked by a negative contrast edge. Space constant of the bars and edge was 1.0 min arc. Data is shown for IRP (open circles) and MEW (filled circles). The subjects indicate a consistent pattern of bias, with the bars being biased away trom the flanking edge at least at the smaller flanking distances tested. This is similar to the data of Badcock and Westheimer (1985a), where they showed that a positive contrast target line is repelled by a negative contrast flanking line.

100 Tj

T3

CL

4

3 5 6 7 8 9 1 0

flanking distance (mln arc)

Figure 3.32: Vernier thresholds are shown for Gaussian bars flanked by an edge, the bars and edge having a space constant of 1.0 min arc. The bars were of positive contrast polarity and the flanking edge of negative contrast polarity. (Open circles: IRP, filled circles: MEW).

Chapter 3: Spatial interference with relative localisation

-5 -

-10- -15

6 8 1 0 12 1 4 16

flanking distance (mln arc)

Figure 333: Location biases for localising Gausian bars with a single edge flank. Above is shown pooled data from two observers, IRP and MEW. The open circles represent data from the condition with a positive contrast polarity of edge flank, and the filled circles the data from the condition with a negative contrast polarity edge. Space constant of the bar targets and edge flank was 2.0 min arc. The data indicate that a positive contrast bar is attracted to an edge of the same contrast, whereas it is repelled by an edge of different contrast.

100 3

13

CL

6 8 1 0 12 1 4 1 6

flanking distance (min arc)

Figure 3.34: Vernier thresholds for 1.0 min arc Gaussian bars with a flanking edge with the same space constant are shown pooled across the two subjects (MEW and IRP); the open circles are for the positive contrast flank condtion, and the filled circles for the negative contrast flank condition.

Chapter 3: Spatial interference with relative localisation

20 -1

-10

1 0 1 5 20 25

flanking distance (min arc)

Figure 335: The data above is pooled across the two observers (IRP and MEW). The open circles are for the condition where the flank is of positive contrast, and the filled circles are for the negative contrast flank condition. The space constant of the

bars was 4.0 min arc.

100 q

a.

1 0 1 2 1 4 1 6 1 8 20 22

flanking distance (min arc)

Figure 3.36: Vernier thresholds for localisation of Gaussian bars are shown pooled across the two subjects. The open circles represent data from the positive contrast flank condition ,and the filled circles data from the negative contrast flank condition.

_ _

Chapter 3: Spatial interference with relative localisation

Mean shift values obtained are presented in figures 3.31-3.36 as a function of the distance between the centre of the cumulative Gaussian flanking edge and the Gaussian target bar. Results of the observers are similar. The perceptual location of a blurred bar is influenced by the presence of a nearby blurred edge. However, the effects are smaller than those reported in the first two experiments, and also smaller than observed when the target feature is an edge. This is of course partly due to the use of only one flanking feature in this experiment, which would roughly halve the size of the induced shift in location of the target features. Both observers were also of the opinion that it was easy to localise a bright bar in the presence of an adjacent edge, and the small size of the mean-shifts obtained may reflect this phenomenal impression.

Attraction is observed for small separation of the target bars and a bright flanking edge, confirming an earlier study which had broadly similar stimulus configurations (Ganz 1964). Perceptual repulsion of the target bar and flank edge was observed at small separations when the flanking feature is a dark edge. The general pattern is similar to bar-bar interactions, although it appears that repulsion is not present at greater flanking distances for either polarity of flank. The data is not firm enough to support a stronger conclusion.

92

Chapter 3: Spatial interference with relative localisation

3.4.5: Experiment 5: spatial interference for edge targets with edge flank

80 -I 60 - 40 “ 20 - -20 3 4 5 6 7 8 9 1 0

flanking distance (min arc)

Figure 3.37: Mean shifts are shown for edge targets, with a single cumulative Gaussian edge flank, which varied between flanking the upper and lower edge. Data is shown for two subjects (GLC; open circles and FMRM; filled circles). Space constant of the edges was 1.0 min arc. The flanking edge had a contrast 50% that of the target edges and had a negative polarity- i.e. the appearance of the target was of a dark region in one quadrant, beside two brighter quadrants. Attraction is evident at small flanking distances for both subjects despite the polarity of the target feature, echoing

the results where the flank feature was a bar.

_ « T3

I

fianking distance (min arc)

Figure 3.38: Localisation thresholds are shown for edge targets, with an edge flank; the flanking edge had a contrast 50% that of the target edge and a negative polarity. Data is shown for two subjects (GLC; open circles, and FMRM; closed circles). Space

Chapter 3: Spatial interference with relative localisation