What is white

1
What is white?

• Dirk Beer
• Ana Dinca
• Donald MacLeod
• U. C. San Diego

2
ltexample stimulus on this pagegt
3
ltexample stimulus on this pagegt
4
ltexample stimulus on this pagegt
5
Absolute White Settings (VSS 2005)
b S/(LM)
Blue
EEW
Red
r L/(LM)
Green
Yellow
6
Variance is greatest in the yellow-blue direction.
b S/(LM)
Blue
Red
r L/(LM)
Green
Yellow

• Similar to discrimination (discriminate from
  internal standard)
• MacAdam (Le Grand, 1949) (colorimetry)
• JND contour from white (Highnote, 2004)

7
Compare to existing data
8
Compare to existing data
alpha
White settings Beer MacLeod (2005) 19 subjects,
50 setting each
1.7 /- 0.1
Color matching ellipses MacAdam, Wyszecki
Fielder Nagy, Eskew, Boyton (1986) 6 subjects,
52 white ellipses
1.2 /- 0.1
Why are white settings much more variable in the
non-cardinal?
9
Two hypotheses

• 1. Surround vs. No Surround?
• No surround, no implied illuminant
• Randomly chose an implied illuminant from
  distribution of natural illuminants
• 2. Delay forces judgment using non-cardinal
  mechanisms
• Abutting stimuli can be judged by local contrast
• Separated stimuli must be judged using memory

10
Surround vs. No surround

• MacAdam data
• Variance of color matches
• With surround
• 4AFC Discrimination
• No surround

alpha
1.2 /- 0.1
1.3 /- 0.1
Non-cardinal Threshold higher without surround
Note Small non-cardinal orientation accounted
for in colorimetry
11
Delay vs. No delay
alpha
4AFC Discrimination Same-different
1.3 /- 0.1
1.4 /- 0.1
?
Tempoaral Same-different, 1 second delay
1.5 /- 0.1
Different?
?
500ms 1000ms 500ms
Delay seems to increase non-cardinal
threshold Longer delays necessary? Tested 1 sec
- 9 sec delays
12
Alpha (non- cardinal threshold/ variance)
4AFC Same-diff 1sec
1-9sec Matching Discrimination
Same-different Setting
Experiment
13
Conclusions

• Judgements without a surround seem to have
  increased variance in blue-yellow unique hue
  direction
• Subject forced to use internal guess about
  illumination?
• Delay between stimuli leads to greater
  non-cardinal variance.
• Infinite time delay must result in absolute
  judgement
• 10 second delay is not yet near absolute judgement

14
Implications

• There is a real, and theoretically important
  elevation of threshold in the unique blue-yellow
  direction
• non-cardinal
• holds in different experiments
• not well explained (despite our efforts)
• Large, unexplained non-cardinal variance of white
  settings
• Longer delays between stimuli?
• Color-difference metrics may be inadequate
  for absolute color judgements.

15
b S/(LM)
Blue
Red
r L/(LM)
Green
Yellow
Alpha
1
1
16
(No Transcript)