Stereoscopic Size Constancy Lab



  1. Purpose and Goals
    1. To determine if size constancy depends upon the number of depth cues, i.e., the richness of the depth information.
    2. To test for the limits of stereopsis.
    3. The develop an understanding of the difference between quantitative and qualitative predictions
    4. To compare data to hypothesis when the hypothesis is mathematically precise
    5. To learn more about designing your own experiment
  2. Size Constancy - Briefly, will discuss more in class next week
    1. Observations
      1. Objects have smaller retinal image sizes the farther away they are from the image.
      2. But they do not appear to shrink as they recede into the distance.
      3. Due to depth cues. 
    2. Stereopsis:
      1. The measure of the difference of position that images fall on the two eyes is disparity.
      2. The ability to see in depth using the depth cue binocular disparity is called stereopsis.
    3. Question:
      1. Stereopsis (binocular disparity) has a very precise relationship with depth, see below, can it support size constancy when all other depth cues are removed.
      2. How does size constancy change as we add more depth cues.
  3. Quantitative Predictions
    1. A qualitative prediction deals with trends.
      1. An example would be from the retinal acuity lab
      2. acuity gets worse farther out in the periphery. 
      3. This statement does not predict how worse.  Thus, it is qualitative.
    2. A quantitative prediction deals with both trends, but the size of the trends.
      1. To apply this to the last lab it might say, for each degree of visual angle out from the fovea acuity gets worse by 10%.
      2. This type of statement allows a much finer test of the idea.
      3. This type of statement can also be written as an equation

        Acuity = (distance from Fovea * .10*acuity at fovea)+

                         acuity at fovea.

        This is simply a translation of the statement at 3.b.i into a mathematical expression.

    3. Quantitative Elements of this situation.
      1. Size Constancy

        Sr α 1/dp
        Sp =  k

        Sr is the size of the retinal image

        Sp is the perceived size of the object

        dp is the perceived distance of the object (since in our case the distances will all be virtually created)

        α means proportional to.

        k is a constant.

        To translate all this into English: The retinal image size decrease as distance increases such that they are proportional.  That means as distances doubles, the height of the image in the eye is cut in half.  However, the perceive size stays the same regardless of the change in image distance.

      2. Perceived Distance

        Distance from Screen = Viewing dist(mm) * disp(mm)
                                                 Interpup dist (mm)

        Perceived Dist = Viewing Dist - Dist from Screen

      3. These will be brought together when the experiment is explained

The Experiment:

  1. Equipment
    1. Here is the link to the lab:
    2. Special hardware
      1. Filter glasses, red filter over the right eye and the blue filter of the left eye.
      2. Right eye see what is drawn in red, left eye sees what is drawn in blue.
      3. Allows different images to be seen by each eye, which allows for disparity
  2. Design:
    1. Arrangement of Stimuli:

             Standard               Comparison
    2. Left circle will have a disparity of 0. 
    3. There will be no fixation so the horoptor will vary.  You will be measuring relative disparity from this point.
    4. Right square will be a different disparities starting at +10 and moving backwards.
    5. IV1 is disparity of comparison square
    6. IV2 is the presence of depth cues: disparity alone, disparity plus relative height, disparity plus relative height plus texture gradient.
    7. DV is size of comparison square
  3. Method:
    1. Viewing Distance is 75 centimeters (be precise).
    2. Open the lab webpage.
    3. Stimuli Setup:
      1. Disparity of Standard Square = 0 pixels (measure in mm)
      2. Disparity of Comparison Square (in pixels, measure in mm)

        +10, +5, 0, -5, -10.

      3. Make sure atmospheric perspective is unclicked.
      4. First time through: do not use relative height or texture gradient. Make sure they are unclicked
      5. Second time through adjust relative height along with disparity according to the following table (so click relative height):
        Disparity Relative Height
        10 0
        5 25
        0 50
        -5 75
        -10 100
      6. Third time through, adjust disparity and relative height as above and add texture gradient (so make sure these three cues are clicked)
        Disparity Relative Height Texture Gradient
        10 0 0
        5 25 25
        0 50 50
        -5 75 75
        -10 100 100
    4. Pick your method and set the parameters
      1. Choices are:
        1. Method of Constant Stimuli
        2. Method of Limits
        3. Method of Adjustment
        4. Suggestion: try these out to see how they work
      2. Set the parameters for the method as you think will give you sufficient data and clean results
        1. Again, try some of the settings out
      3. Messy results may be easy to collect but no fun to interpret
      4. A new measure.
        1. We are trying to find the size of the comparison that appears to be the same size as the standard circle
        2. This measure is the Point of Subjective Equality (PSE)
  4. Procedure:
    1. After setting up the condition, put on glasses
    2. On each trial respond as directed on the screen.
  5. Generating Predictions:
    1. Measure Perceived Distance for each disparity condition: e.g., viewing distance = 750 mm, disparity = 3.4 mm, interpupillary distance = 63 mm
      1. Distance from Screen = (750*3.4)/63 = 40.5 mm
      2. Perceived Distance = 750-40.5 = 709.5
    2. According to Size Constancy: when two objects appear the same size the retinal image sizes are inversely proportional to the two distance.
      1. in English, an object that is twice as far away has to be half as tall on the retina to appear the same size.
      2. So, the comparison is 709.5/750 = 0.95 as far away
      3. The predicted size is the (standard size)/(rel dist) =
        20 mm/0.95 = 21.1 mm
  6. Write-up: (Full report)
    1. Week 2: Do a graph of your predictions to hand in so I can inspect them.
    2. At end of lab, do all sections of the lab report specificed in the lab report format.
      1. This lab report has a particularly complicated stimulus section you need to measure the following elements:
        • The disparity in mm: from edge or color (red or blue) to edge of magenta on same side
        • The size of the standard in mm
        • the sizes of the comparison in mm if using method of limits or method of constant stimuli
        • The positions on the screen of the two circles (to the center of the circles) from the left side and bottom of screen
        • This last value will change when relative height is changed and all values need to be measured
        • The texture gradient needs to be described.
        • The vertical back shoule be measured for size and position like the circles.
        • The position of the front corners of the texture gradient should be noted.