Color Selection and Scaling

The program cds has been developed for detection and scaling experiments related to color vision. The subject is shown color patches and has to select one of them or has to detect a color patch in a background and a distractor set. Adjustment and adaptive procedures also are possible. This may be used to find unique hues or to run threshold experiments.

Program Features

Color Stimulus Definition

There are 5 color stimuli available. They are called a, b, c, d, and t. The string positions controls which of these five stimuli are visible and where the stimulus' position is. The color coordinates of a, b, c and d have to be specified in the parameter file. The coordinates of t usually are computed as a convex mixture of a and b. The equation is t = w * a + (1.0 - w) * b, where 0 <=w <=1 has to hold. The coordinates of a are given by the parameters ax, ay, and aL, those of b by bx, by, and bL. The factor w is defined by the parameter aweight. Special values for aweight may be used to make t identical to a or b. The parameter aweight may also be used as an adaptive variable in adaptive procedures.

The stimuli a and b must be defined, but they need not be visible. Even t need not be visible, although it is always computed and stored into the data file. Its coordinates are tx, ty, and tL. The coordinates of c and d are cx, cy, cL, dx, dy, and dL respectively.

Stimulus Positions

The stimulus patches are numbered clockwise and their centers are positioned equally distant on a circle around the center of the screen. The circle's radius is patchdist. The position of the first patch on the circle is given by firstangle. This parameter gives the angle between the horizontal and a line from the screen center to the patch. The patches are circular, square or rectangular with their size defined by diameter. The form is defined by patchtype. If the flag L is set then the patches are positioned on a horizontal line, not on a circle and patchdist is the distance between two patch centers.

The number and sequence of stimulus patches is defined by the string positions. The string contains one of the letters "a", "b", "c", "d", or "t" for each position. The number of letters gives the number of patches and the letter gives the name of the stimulus which is mapped to the respective position. Patches are ordered from left to right according to their names in positions if the flag L is set.

If the flag T is set then the position of t is selected randomly. The string positions need not contain the letter "t" in this case. The actual position of the test stimulus t is stored in the parameter target.

Adaptation Field

The adaptation field has width fieldwidth and height fieldheight. Its color is defined by fieldcolx, fieldcoly, and fieldlum. If the flag R is set then the adaptation field is rectangular, otherwise it is elliptical. A fixation mark of size fixmark and luminance fixmarklum may be shown at the center of the adaptation field. If fixmark is 0 then there is no fixation mark. If the flag S is set then the adaptation background for the patch areas is controlled separately from the rest of the adaptation field. Its color is then defined by patchx, patchy, and patchL,

Responses and Response Keys

The subject's response usually is to select one of the color patches. The mapping of response keys to stimulus patches is done via the array keys. It contains the response key numbers which are associated with the color patches as they are ordered in the string positions. Thus keys must have the same number of entries as positions. To make patches unselectable one can set the corresponding key number to numbers which cannot be generated by response keys, like -1. The following example defines four stimulus patches, the target stimulus t is positioned randomly and may be selected by one of the keys labeled "4", "5", "2", and "1" in the separate numeric section of the keyboard. Remember that the ASCII code of those numeric keys is 48 plus the respective number.
   positions = "bbbb"                /* 4 patches filled with color b */
   firstangle = 1350                 /* top left patch is 1st */
   flags = "TRl"                     /* random target position, 
                                        rectangular background field */
   switchtype = 0                    /* keyboard response */
   keys = [48+4, 48+5, 48+2, 48+1]   /* keys are: 4, 5, 2, 1 */
   evaluation = 2                    /* evaluate whether target detection */
Every stimulus patch not containing the target t contains stimulus b. The arrangement is such that the first stimulus is the top left patch at 135deg. The stimulus field has two lines and two columns.

Response Evaluation

The parameter evaluation determines how responses are evaluated. The result of the evaluation is stored in response and the response time is stored in rtime. The following types of evaluation are possible:

evaluation = 0. No special evaluation is done and response is set to the position which corresponds to the response key.

evaluation = 1. The response is coded as a yes/no response. "yes" corresponds to the selection of a or d and "no" to the selection of b or c. This is useful for adaptive designs where "yes" leads to an increase of the adaptive variable and "no" to a decrease. If aweight is the adaptive variable then a decrease of aweight reduces the similarity of the target stimulus t to a.

evaluation = 2. This also leads to yes/no coding of the response. The response is "yes" if the target t is selected and "no" if any other position is selected. This is useful for detection tasks where the subject has to find the target stimulus among several distractors.

Random Intensity Variation

Some experiments require a random fluctuation of distractor stimuli. An example is a detection experiment where the distractors vary in brightness and the task is to detect one deviant color among the distractors. Such a random variation of stimulus intensity is done when the flag C is set. If this is the case then fluctuation is half of the luminance range for the distractor stimuli. Thus if C is set then the effective luminance of the distractor stimuli (this are all displayed stimuli not equal to a, b, and t) is in the interval (1.0 +- fluctuation) * L, where L is the luminance given in the parameter file.

Sound Cue

For detection experiments it may be necessary to present a sound cue while the target stimulus is shown in order to indicate the presentation interval. This is done if soundcue is positive. It gives the cue's duration. cuepitch gives its frequency and aftercue is the delay between the cue off and stimulus on event.

Adjustment Methods

There are two different adjustment procedures available to set the target color t. If the flag A is set then the color of t may be adjusted by mouse movements. With no buttons pressed the mouse moves in the CIE xy-chromaticity diagram and the color coordinates of t are set accordingly. When a single button is pressed, then mouse motion controls the luminance of t while chromaticity is kept constant.

If the flag M is set then t also is adjusted by mouse movements, however, motion in color space is restricted to the line connecting colors a and b. In this case the parameter actually adjusted is the weight w in the mixture \eqrefcdsmix. Pressing the stopkey signals the program that adjustment is finished.

Program Parameters

aftercue (int)
Pause between end of sound cue and beginning of the patch display.

aweight (float)
Weight of the stimulus a for computing the target stimulus t = aweight*a + (1-aweight)*b.

ax , ay , aL , bx , by , bL , cx , cy , cL , dx , dy , dL (float)
Color coordinates of the stimuli a, b, c, and d respectively.

colortable (float)
This is a table of xyL-colors. The colors may be referenced by character entries in the string positions. Character 0 corresponds to the first color in colortable, charcater 1 to the second ... Thus the character ASCII codes are index values for entries in the color table. This feature may be used to assig arbitray colors to multiple stimulus positions.

cuepitch (int)
Frequency [Hz] of the sound cue.

diameter (int)
Defines color patch size. It is the diameter of the circular patches and the length of one side for square patches. Nonsquare rectangular patches have one side equal to diameter and the other equal to diameter/3.

dsptime (int)
Stimulus patch display duration. Adjustment methods are run without time limit.

evaluation (int)
Selects the type of response evaluation:
0
No special evaluation. Set response to the position of the response key given in the array keys.
1
Recode the response to yes/no, where "yes" corresponds to the selection of a or d and "no" to the selection of b or c.
2
Set response to "yes" if the target t is selected and to "no" otherwise.
fieldheight , fieldwidth (int)
Height and width of the adaptation field.

firstangle (int)
The angle between the horizontal and the connection between the screen center and the center of the first patch.

fixmark (int)
Size of the fixation mark. A 0 size means that there is no mark.

fixmarklum (float)
Liminance of the fixation mark.

flags (string)
These are the possible flags:

A
Use a 3-dimensional adjustment method to adjust the target color t. In this case t is not computed as a mixture from a and b.

C
If this flag is ON then the luminance of the distractor stimuli fluctuates randomly.

D
Do not mix target color from a and b but set t directly from coordinates given in the parameter file or adjusted by the adjustment method.

G
Separate stimulus set into two groups and show them next to each other.

L
Use horizontal linear positions for the patches instead of circular ones. Color patches are ordered from left to right according to their names in positions.

O
Use an oval adaptation field instead of a rectangular field.

P
Repaint the display for each trial.

S
Allow definition of a special adaptation color for the color patch fields.

T
Randomly position the target stimulus t at any of the positions in positions. The actual position is stored in target.
fluctuation (float)
Gives the random fluctuation range of the distractor luminance values if the flag C is set. If C is set then distractor luminances are in the interval (1.0 +- fluctuation) * L, where L is the luminance given in the parameter file.

keys (int)
Gives the numbers of the possible response keys in a sequence which simultanously defines how the keys are associated with positions. For keyboard keys the numbers correspond to the ASCII codes. For switches the numbers correspond to the standard PXL switch numbers.

patchdist (int)
For circular patch positions this is the radius of the circle where the patches are positioned on. The center of the circle always is the center of the screen. The patch centers lie on the circle. If the flag L is set then the patches are positioned on a horizontal line. In this case patchdist is the distance between centers of adjacent patches.

patchtype (int)
Form of patches: 0 = circular disks, 1 = horizontal bars, 2 = vertical bars, 3 = square bars.

patchx , patchy , patchL (float)
Color patch area adaptation color. Used only if flag S is set. If the flag is clear then the color patch area color is identical to the adaptation field color.

positions (string)
Defines the number of stimulus patches and which stimuli are shown at the possible positions. Each character in positions is one of "a", "b", "c", "d", and "t" which corresponds to the five possible color stimuli a, b, c, d, and the target t. If the flag T is set then t is positioned randomly. The serial position of each patch in this array also establishes the correspondence between the patches and the response keys.

soundcue (int)
Duration of sound cue. No sound cue is given if soundcue is 0.

target (int)
The position number of the target stimulus t. Numbers start with 1 corresponding to the position at firstangle and run clockwise.

tx , ty , tL (float)
Color coordinates of the target stimulus t. Note that t always is a convex mixture of a and b.

Example Experiments

Color Discrimination

There are 4 color patches where three of them always contain the same color (b) and one of them contains the target t. The subject't task is to detect t among the 3 distractors. The distractor color is white/gray and the target is slightly colored. It actually is a mixture of white (b) and some highly saturated color which in color space lies on one of the confusion lines for color blind persons. Thus the experiment is used to detect color vision anomalies by determining the discrimination ability along confusion lines in color space. The distractors' luminance values are randomly fluctuated such the the subject has to discriminate between luminance and hue variation.


Parameter file cblind.x from directory \pxl\app\cds

Unique Hues

This experiment determines unique hues by an adaptive procedure. There are three linear positioned patches in each display. The outer two colors are similar to opposite uniqe hues like yellow on the left side and blue on the right side and the center is some target color. The subject's task is to chose either a "left" or a "right" key according to whether the center color either contains more of the left or more of the right color. The adaptive procedure will converge to a color which neither contains the left nor the right color. The center color is a convex mixture of stimuli a and b which are not displayed. The stimuli c and d only tell the subject which colors to look at.


Parameter file uniq.x from directory \pxl\app\cds

Adjustment

This example creates an adjustment procedure where there is a single color patch on the screen whose color can be adjusted by mouse movements.


Parameter file adjcol.x from directory \pxl\app\cds

Show color patches and get a choice or rating response

Stimuli are 6 color patches or 2 groups of 3 patches on a bright background field. The task is a rating response or a binary choice task. The keyboard is used for rating and the mouse buttons are used for choice.


Parameter file pref2.x from directory \pxl\app\cds

Back to table of contents


Author: Hans Irtel

irtel@psychologie.uni-mannheim.de