Color tuning in metacontrast masking

1
Color tuning in metacontrast masking

• Dirk Beer and Don MacLeod
• University of California at San Diego
• La Jolla, CA

2

• How is color represented in the cortex?
• Retina, lateral geniculate nucleus Luminance and
  two chromatic cardinal axes (Encoding by set of
  three parallel signals).
• Cortex No clear organization, no privileged
  axes?
• Metacontrast masking
• Yellott Wandell (1976) color selective
  masking occurs dichoptically
• Anstis Becker (UCI/OSA 2001) luminance
  polarity selectivity.
• Beer MacLeod (VSS 2002) masking is
  color-polarity selective.

3
Metacontrast Masking
time
33 ms
50 ms
33 ms
Click
Saw It? Yes/No
SOA 83 ms
Yes/No, 127 conditions, randomized QUEST
staircases, 6 naïve subjects, 8
hours/subject Display 17 x 13 deg Test
diameter 1.5 deg Ring diameter 3.0
deg Luminance (background) 4000 cd/m2
4
Extended Maxwellian View (BIGMAXTM)
High luminance 100, 000 cd/m2
5
Improved gamut with additional dichroic filters
100
10
S/(LM)
1.0
0.1
0.01
0.5
0.6
0.7
0.8
0.9
1
L/(LM)
Max luminance with filters 20,000 cd/m2
6
Preliminary experiment determining SOA
Use 83 ms onset asynchrony
7
Demo
8
Threshold elevation plot (luminance/red-green
plane)
CLM
15
LIGHT

10
5
RED
GREEN
0
-5
-10
DARK
-15
-15
-10
-5
0
5
10
15
CL/(LM)
9
Predictions Cardinal axis model with adaptation
10
Luminance - Red/Green Plane (N3)
CLM
20
15
10
5
0
-5
-10
-15
-20
-20
-10
0
10
20
CL/(LM)
DARK
11
Luminance Red/Green Plane (N3)
12
Threshold elevation plot (isoluminant plane)
CS/(LM)
BLUE

RED
GREEN
YELLOW
CL/(LM)
13
Red/Green Blue/Yellow Plane (N3)
14
Results Conclusions

• Polarity selectivity Color masking is polarity
  selective, luminance
• masking less so.
• Anisotropy
• - Color (L-M cone axis) masks luminance,
  luminance does
• not mask color
• - Color (L-M cone axis) masks color (S-cone
  axis),
• but S-cone axis does not mask L-M cone axis.
• Tests that escape masking are seen with roughly
  veridical luminance and color
• contrast, inconsistent with a model of
  sensitivity change of cardinal
• mechanisms.
• Inconsistent with a simple cardinal mechanisms
  model
• - similar color difference can have different
  masking effect
• - tuning is not broader for intermediate
  directions
• - masking does not simply reduce contrast

15
Test that escape masking are matched with
veridical luminance contrast.
16
Questions

• Why less luminance polarity selectivity than in
  previous ?
• Higher mean intensity?
• Do S-cone spots mask luminance, as L-M cone
  spots do?