Dc. Knill, IDEAL OBSERVER PERTURBATION ANALYSIS REVEALS HUMAN STRATEGIES FOR INFERRING SURFACE ORIENTATION FROM TEXTURE, Vision research (Oxford), 38(17), 1998, pp. 2635-2656
Optical texture patterns contain three quasi-independent cues to plana
r surface orientation: perspective scaling, projective foreshortening
and density. The purpose of this work was to estimate the perceptual w
eights assigned to these texture cues for discriminating surface orien
tation and to measure the visual system's reliance on an isotropy assu
mption in interpreting foreshortening information. A novel analytical
technique is introduced which takes advantage of the natural cue pertu
rbations inherent in stochastic texture stimuli to estimate cue weight
s and measure the influence of an isotropy assumption. Ideal observers
were derived which compute the exact information content of the diffe
rent texture cues in the stimuli used in the experiments and which eit
her did or did not rely on an assumption of surface texture isotropy.
Simulations of the ideal observers using the same stimuli shown to sub
jects in a slant discrimination task provided trial-by-trial estimates
of the natural cue perturbations which were inherent in the stimuli.
By back-correlating subjects' judgements with the different ideal obse
rver estimates, we were able to estimate both the weights given to eac
h cue by subjects and the strength of subjects' prior assumptions of i
sotropy. In all of the conditions tested, we found that subjects relie
d primarily on the foreshortening cue. A small, but significant weight
was given to scaling information and no significant weight was given
to density information. In conditions in which the surface textures de
viated from isotropy by random amounts from stimulus to stimulus, subj
ect judgements correlated well with the estimates of an ideal observer
which incorrectly assumed surface texture isotropy. This correlation
was not complete, however, suggesting that a soft form of the isotropy
constraint was used. Moreover, the correlation was significantly lowe
r for textures containing higher-order information about surface orien
tation (skew of rectangular texture elements). The results of the anal
ysis clearly implicate texture foreshortening as a primary cue for per
ceiving surface slant from texture and suggest that the visual system
incorporates a strong, though not complete, bias to interpret surface
textures as isotropic in its inference of surface slant from texture.
They further suggest that local texture skew, when available in an ima
ge, contributes significantly to perceptual estimates of surface orien
tation. (C) 1998 Elsevier Science Ltd. All rights reserved.