Perception

A computer graphic method for extracting a natural image of an individual's facial prototype, or average appearance, from a number of different images of that individual is presented. The process improves upon previous photographic and computational techniques. Synthesis of a person's average expression and pose from a sample of images is derived in an automatic and quantitative way. Possible uses of composite faces produced in this manner in psychological investigations of facial qualities (eg attractiveness) and in applied areas such as telecommunication are pointed out.

Principles of perceptual organization can cause simultaneous lightness contrast to be seen in patterns in which the inducing regions are not proximal to the target. In three experiments a match was made between two grey targets, one of which belonged to a white perceptual grouping and the other to a black one. In experiment 1, perceived belongingness was a product of common fate and figural alignment (columns of black dots and columns of white dots moved up and down in counterphase). In experiment 2, belongingness was evoked with only figural alignment. In experiment 3, belongingness was induced by common fate alone. In all three experiments it was found that the grey target belonging to the white group appeared darker than the grey target belonging to the black group.

Orientation illusions induced by two-dimensional stimuli, such as square outline frames or plaids, have been more or less adequately accounted for in terms of repulsion of the vertical test stimulus from the axis of symmetry nearest vertical of the inducing stimulus, whether that axis is real or virtual. Recently, data have been obtained which directly suggest a more complex mechanism: one in which the observed illusion is the sum of all effects—complementary and antagonistic—induced by all axes flanking vertical which are sufficiently close to vertical to exert a significant effect. Experiments are reported in which this latter hypothesis was directly tested by using nonorthogonal plaid component gratings and varying the real-axis orientations while a virtual plaid axis remained fixed in orientation at 10° from vertical. The data indicate that the real component gratings modulate the virtual-axis effect.

Moiré patterns can produce striking movement effects and in more complex stimuli can induce vivid stereoscopic depth. The physical rules underlying these phenomena are reviewed and their relationship to psychophysics is discussed.

First, it is shown how moirés in ‘optical line interference’ patterns are created by superimposing periodic visual stimuli, eg gratings, and shifting them relative to each other. When two gratings are presented in this manner, small differences in spatial frequency, orientation, and speed are magnified. This magnification has prompted the use of moiré patterns both in industry and in art where their enhanced sensitivity to misalignment and spatial distortion has been widely exploited.

Next, it is demonstrated how enhanced depth in ‘stereoscopic interference’ patterns is produced by presenting grating stimuli in two (or more) depth planes. The perceived depth effect in the resulting moiré pattern can be elicited similarly by binocular disparity and motion parallax.

Finally, it is described how perceived movements occurring in different directions and at different depths are the basis for the perceptual ‘irritations’ that fascinate observers in complex moiré patterns. The use of moirés for the noninvasive examination of the human retina by aliasing is discussed.

In ‘left’ neglect—a frequent consequence of damage to posterior areas of the right cerebral hemisphere—the relational term ‘left’ immediately raises the question ‘left of what?’. It is shown that on a simple copying task the answer can be ‘left of the principal axis of the object or objects depicted’. Under conditions in which the stimulus material was intrinsically controlled for lateral extent and position, ‘object-centred’ neglect seemed to reflect some basic principles of perceptual parsing.

A previously unreported size illusion in some commonly used stimuli—a large (global) letter made up of small (local) letters (eg a large H made up of small Es) is described: the size of the local letters forming the horizontal component of the global letter appears to be overestimated. Global and local letters made up of horizontal and vertical components (E, F, T, and H) were employed as stimulus material. The first experiment showed that the illusion was consistently perceived, and was present for most of the stimuli tested; ie for all global letters and for all local letters except H. A second experiment showed that the size distortion is not specific to a particular dimension: instructing observers to focus on the height or the width of the local letters did not affect the extent of the illusion. The effect of some stimulus manipulations was also investigated.

The three-dimensional (3-D) apparent motion of lines, outline triangles, and light points was studied in four experiments. The stimulus sequences were beginning and end patterns of 3-D motions of a line and a triangle. Light-point patterns corresponded to the ends of the lines and the vertices of the triangles. Perceived motion of lines and light-point pairs resembled the distal motions that were used to construct the proximal patterns. The correspondence was striking for configurations that appeared to move in depth. Outline triangles and light-point triplets produced a strong correspondence between distal and perceived motions when the three sides appeared to be translating in depth. The correspondence was reasonably good for the other motion patterns when scoring included an appropriate second category. The results support the conception of structural or internalized constraints: light points were processed as if they were connected (unity constraint) and proximal change in linear size (or distance between light points) was perceived as rigid 3-D motion (rigidity constraint).

Although the observer's motion can elicit perception of relative depth, it is less successful in doing so when competing pictorial information is available. However, the evidence for this may be affected by limited extents of motion and by equidistance tendencies. Results obtained when monocular observers judged the orientation-in-depth of trapezoidal and of rectangular surfaces, during lateral head motion of extents 0 cm to 30 cm, are described. When the motion extent was less than 30 cm, trapezoidal surfaces were misperceived because they were interpreted as rectangular; this pictorial information was overriden only when the motion extent was 30 cm. The results may reflect the sequential nature of motion information and the redundancy of information in normal viewing: pictorial information may take precedence when motion is limited, but motion information can be indefinitely augmented. Comments are directed to (i) the use of Ames ‘distorted rooms’ in this area of research, and (ii) the ‘ecological’ interpretation of pictorial information.

A stimulus is described that demonstrates the spatial pooling of colour information in the visual system. Chequerboards (or gratings) consisting of alternating squares (or stripes) of complementary colours become achromatic at particular spatial scales; such stimuli have been named ‘transchromatic’ stimuli. Colour pools are much larger than the receptive fields that respond to luminance contrast. Some measurements are described which form the basis for estimates of the size of the colour pools. The size of colour pools varies according to the colours involved. For red—cyan and green—magenta complementary pairs colour is pooled at spatial frequencies above about 7–8 cycles deg⁻¹, implying pools whose diameter is around 8 min arc. For yellow—blue complementary pairs the corresponding figures are about 4 cycles deg⁻¹ and 15 min arc. Some phenomena of normal colour vision, colour blindness, and the development of infant vision are discussed in the light of these findings.

Neon colour spreading has been shown to disappear if certain figural conditions are not met. Evidence is presented which suggests that these conditions are only incidentally related to the neon spreading effect; in particular, that they can be violated as long as the structure remains compatible with the interpretation of a transparent surface. It is proposed that neon spreading and classical colour assimilation share the same basic mechanism, and that the peculiar perceptual attributes of the former derive from the perceptual scissioning of ordinary assimilation colour. This process is identical to that occurring with nonillusory colours in phenomenal transparency.

Two experiments were carried out to investigate the perception of complex auditory-spatial patterns. Subjects were asked to identify alphanumeric characters whose patterns could be outlined acoustically through the sequential activation of specific units in a speaker array. Signal bandwidths were varied systematically in both experiments. Signals in experiment 1 had sharp onsets and offsets; envelope shapes in experiment 2 were much more gradual. Subjects showed considerable ability in recognizing alphanumeric patterns traced with signals of varying acoustical composition. Reductions in the steepness of signal attack and decay produced limited declines in pattern recognition ability. Systematic trends in the relation between patterns and the distribution of incorrect responses suggest that subjects performed a pattern-matching task, in which identifications were made on the basis of component features. The unexpected pattern recognition abilities that subjects demonstrated in both experiments suggest that spatial hearing, like vision, has access to mechanisms for amodal spatial representations.