Perception

When a black and a white square on a grey surround exchange places, it was previously shown that on a dark surround it is the white square, and on a light surround it is the black square, that is seen in apparent motion (AM). Thus the higher-contrast square carries the AM. We now show that the same is true for second-order AM of texture-defined squares. Squares were defined by four different textures: by anisotropy (horizontal versus vertical random dashes), by alphanumeric letters, by hash marks, or by dot size. The result was that the square that differed more from the surround in texture properties carried the second-order AM. Judgments of texture salience revealed a high correlation between salience and apparent motion. In a third experiment, crossover AM between dissimilar textures was investigated, and it was found that the more salient textures carried the AM. Results cannot be explained by the concept of ‘texture activity’, but instead indicate that the system extracts a measure of ‘texture contrast’ prior to analysis of salience and apparent motion.

The stimuli in these experiments are square-wave luminance gratings with an array of small random dots covering the high-luminance regions. Owing to the texture, the direction of these gratings, when seen through a circular aperture, is disambiguated because the visual system is provided with an unambiguous motion energy. Thus, the direction of textured gratings can be varied independently of grating orientation. When subjects are required to judge the direction of textured gratings moving obliquely relative to their orientation, they can do so accurately (experiment 1). This is of interest because most studies of one-dimensional motion perception have involved (textureless) luminance-defined sine-wave or square-wave gratings, and the perceived direction of these gratings is constrained by the aperture problem to be orthogonal to their orientation. Thus, direction and orientation have often been confounded. Interestingly, when subjects are required to judge the direction of an obliquely moving textured grating during a period of adaptation and then the direction of the motion aftereffect (MAE) immediately following adaptation (experiments 2 and 3), these directions are not directly opposite each other. MAE directions were always more orthogonal to the orientation of the adapting grating than the corresponding direction judgments during adaptation (by as much as 25°). These results are not readily explained by conventional MAE models and possible accounts are considered.

In a geometrical figure in which long vertical lines are each crossed by a series of short oblique lines, an illusory effect is obtained such that the orientations of the long lines are perceived as nonvertical and shifted away from the orientation of the oblique lines (the Zöllner illusion). In addition, the vertical separation between the crossing (oblique) lines is perceived as less than that if the crossing lines are horizontal (the Judd illusion). It has previously been shown that these two effects are closely related, and a single-process account has been proposed in which both effects are explained by a computational model involving band-pass spatial filtering of the figure by means of difference-of-Gaussians (DOG) filters. Two arguments are presented against the latter account. First, in an opposite-contrast-polarity figure with, for example, white vertical lines and black crossing lines on a mid-grey background, the peaks in the DOG filter output are such as to predict the reversal of the Zöllner—Judd effects. It is shown by demonstration that this prediction is disconfirmed, and that the normal effects are obtained. Second, it is shown that the normal Zöllner—Judd effects are obtained in the absence of the long vertical lines, and in the presence of anomalous contours. The latter effects are also in contradiction to the band-pass-filtering model. These findings are discussed in relation to a dual-process account of the Zöllner—Judd effects.

After prolonged viewing of a three-element target in which the middle element is spatially offset, subsequent viewing of the same three elements in alignment results in the middle element appearing to be offset in the opposite direction. This adaptational aftereffect to a spatial offset was investigated with elements which were spatial-frequency narrowband and equidetectable to ascertain (a) the properties of the mechanisms involved and (b) the nature of the underlying computation. Evidence is presented in favour of an orientational-grouping, rather than a purely positional computation, underlying this aftereffect. A dual site of adaptation is proposed: one which receives input from the orientation extracted from the output of linear filters, and another which receives input from the orientation derived from grouping processes working on the contrast-energy representation. These may correspond to the mechanisms which are thought to underlie the processing of real and subjective contours.

This paper examines people's ability to make judgments which require them to know the relative positions of objects that are not simultaneously visible is examined. It has previously been shown that people can accurately perform such a task. The current experiments test the capacity limits for such tasks. Two experiments were conducted that required subjects to make spatial judgments based on sequences of points presented two at a time. It was shown that, whereas subjects can perform accurately when memory for a small number of dots (about four) is required, increasing the number of dots results in a radical reduction in performance. This argues against both the idea that spatial memory is based on a linguistic description and the idea that it is based on an image-like representation. Rather it appears that one can form an accurate representation of the spatial properties of a small number of objects.

The lengths of lines and the sizes of angles were measured in freehand drawings of cubes produced by 190 children and 158 adults. The lengths of oblique lines depicting receding cube edges were foreshortened relative to horizontal lines showing nonreceding edges. In the drawings from children aged 9 and 10 years the obliques were foreshortened by about 40%, compared with 30% in adults' drawings. The amount of foreshortening was not correlated with the angle at which the obliques were drawn. Line lengths were also foreshortened in ‘transparent’ drawings, which are often said to show ‘what is known’ about cube structure. One explanation of the results is that line lengths are foreshortened by children and adults to create the visual impression of equal-length cube edges, even in transparent drawings. This foreshortening fits with how the receding edges of a square face of a cube project to a vantage point. Children and adults who use foreshortening are not depicting the structure of cubes by matching features such as equal-length edges with similar features in the physical lines on the page, as has been suggested in some recent approaches to drawing development.

In the first of three studies, children (aged 8 to 14 years) were found to perform worse than young and middle-aged adults in unprompted identification of odors, with average performance much like that of elderly adults. Comparisons on other tasks, specifically odor threshold, prompted odor identification, and object naming (Boston Naming Test), across the life span (five groups) revealed that children have the same excellent olfactory sensitivity as young adults and merely lack odor-specific knowledge that accumulates slowly through life. Such knowledge apparently accumulates so slowly that age-associated discriminative losses, measurable by early middle age, begin to wear away gains obtained through experience before odors can become overlearned. In the second study, a novel adaptive psychophysical method, the step procedure, confirmed the equivalent sensitivity of children and young adults. In the third study, a paired-associate task illustrated the sluggish course of odor learning. Young adults outperformed children, though the youngest group, first graders, made up ground relatively fast. For children and adults, common odors facilitated performance relative to novel odors. The outcome highlighted the relevance of semantic factors in odor learning irrespective of age.