This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
• retinal disparity + X = (theoretically) absolute/metric distance, where X is:– vergence angle– vertical disparities– other cues….
• “scaled horizontal disparity” - process of combining horizontal retinal disparity with other info. to yield metric distance. This is theoretical possibility; generally human vision system (HVS) seems unable to do this
• adjust depth of hemi-cylinder to make it appear cylindrical– depth range {0,9}cm (recall height is 3cm)– distances {40,50,60,70} cm– diagonal depth profile (left/right) per trial is random– keyboard press increase/decrease depth– initial depth randomly selected– no time limit– 15 settings per distance, 60 total
• sub.’s set significantly more depth for far cylinders than near ones– true individually for all but sub. F & PBH
• dist. effect not accompanied by reduced variability in sub.’s setting (i.e. std. dev.) compared to ACC; this contrasts w/ majority of sub.’s impression that matching task was easier than ACC task
• matching task results consistent with hypothesis that sub.’s set matching cyl. depth based on incorrect estimated of viewing distance—just like in ACC task
• suggests sub.’s do NOT use alternative strategy to set relative shapes of 2 objects using dist. ratios alone (i.e. w/out needing veridical distance scaling)
• the dist. ratio strategy appears unused, yet this didn’t result all sub.’s making similar shape settings ACC and matching tasks– sub.‘s D & F are similar, but not C and PBH
• w/out dist. ratio strategy both reference and match cyl. are subject to mis-scaled dist., this could amplify the error and might explain why matching task showed stronger effect of dist. than ACC task
• Sequence:– sub. press ‘start’ key – cylinder appears and is stationary for 1s– then move 17.8 cm/s and disappears– sub. press key to indicate if shape appeared
to ‘expand’ to ‘contract’ in depth extent• experimenter originally illustrates
expand/contract concept to sub.’s using cardboard model
• for expansion/contraction case {1/4, ½, 1,2,4}, effect of motion direction is significant
• Point of Subjective Equality (PSE) - 50% point of fitted function is the amount cylinder must expand/contract when moving toward/away so that sub. is equally like to judge it expand/contract (i.e. perceived it as physically constant)
• for sub. E point of (PSE) for movement toward sub. is 0.89 and movement away is 1.35; diff. is stat. sign.
• direction of motion has sign. effect consistent across 5 or 7 sub.’s; exceptions:– F no clear effect, PSE’s 1– PBH both PSE’s > 1 (so cyl. must always expands)
• 66% of physically constant cylinders moving towards sub. perceived as expanding; 35% of physically constant cylinders moving aways perceived as expanding
• individual differences in level of expansion/contraction needed with toward/away motion are large, but differences in PSE’s are not stat. sign.; indicates level of expansion for away motion and level of contraction for toward motion is isotropic.
• sub.’s failed to accurately identify objects that exhibited physical shape constancy as they moved in depth
• a cylinder of constant perceived shape had to be physically expanded when moving away and contracted when moving forwards– predicted by [Johnston 1991]
object’s changing retinal image needs to be “scaled” by HVS dist. information to perceived a constant 3D object. What if dist. info. scaling for size constancy is incorrect?
– authors discount this• sub.’s shown physical cylinder model• cylinders in all trials same physical size but result
• demonstration of failure of shape constancy• when stereopsis + motion is theoretically
sufficient to indicate a physically constant shaped stimuli, HVS perceives physically constant stimuli to expand and contract with forwards and backwards movement
• Big Question:– Why, given well-documented biases in
perception, humans perceive the world to be a relatively stable place?
• “The blobs scaled appropriately with distance such that mean screen luminance increased with decreasing distance.” ???
• display not recalculated for diff. IPD’s– cylinders at 40,50,60,70cm– screen at 32.5 cm– what does false eye separation do? – would it confound results if some