Perception

Infants recognise their mother's voice at birth but appear not to recognise visual-only presentations of her face until around 3 months. In a series of experiments visual discrimination by infants aged 1, 3, and 5 months of their mother's and a female stranger's face was investigated in visual-only and visual-plus-speech conditions. In the first experiment these infants' discrimination of mother's and female stranger's faces was measured by their visual-fixation-preference scores. Discrimination was found to be facilitated by the addition of speech information. In experiment 2 naive adults viewed silent videotapes of infants from experiment 1 and judged whether the mother had been presented on the infants' left or right. This added further information to the fixation-preference results of experiment 1: it was found that 1-month-olds discriminate mother's and stranger's face only in the presence of speech information, whereas 3-month-olds also do so in visual-only conditions. In experiments 3 and 4 the relative salience of lip movements and voice information in visual recognition of mother's face was investigated. In experiment 3, no significant differences in infants' visual-fixation-preference scores were obtained. However, in experiment 4 adults' ‘where is mother?’ judgments of videotapes from experiment 3 were found to be more accurate in the voice than in the lip-movements conditions, especially for the 3-month-olds and more accurate when mother rather than stranger was talking. It is concluded that young infants' visual recognition of mother is facilitated by addition of speech information, that it is primarily the voice component of speech that causes this facilitation, and that social discrimination is best indexed by a dependent variable which is sensitive to a range of facial cues provided by infants.

The barber-pole illusion and its influence on plaid perception were investigated in two experiments to test the following expectations: (i) apertures which bias the perception of grating motion in directions consistent with plaid direction will facilitate plaid perception, and (ii) apertures which bias the perception of grating motion in directions inconsistent with plaid direction will disrupt plaid perception. In experiment 1 the barber-pole illusion was measured as a function of grating orientation (20°, 45°, and 70°, clockwise and counterclockwise from horizontal), and aperture shape (vertical, horizontal; at each of three elongations). Barber-pole illusions reported with 45° gratings increased with increased aperture elongation. However, this was not found with 20° and 70° gratings; these were almost always reported as moving in a direction parallel to the side of the aperture with which the gratings formed angles approaching 90°. In experiment 2 this dependence of barber-pole illusions on the relative orientation between gratings and apertures was also evident with 45° gratings in oblique apertures; only oblique directions of grating motion were reported. The influence of the same apertures on the separate contrast thresholds required for initial plaid coherence and initial plaid decomposition was measured. In experiments 1 and 2, coherence thresholds were unaffected by apertures, contrary to expectation (i). However, in both experiments expectation (ii) was confirmed; decomposition thresholds decreased in apertures which biased perceived direction of gratings towards vertical (plaid direction), and increased in apertures which biased grating motion away from vertical. Adaptation of plaid mechanisms during measurement of decomposition thresholds was proposed to explain the discrepancy between coherence and decomposition data. Taken together, the results were interpreted as reflecting interactions between mechanisms mediating the barber-pole illusion and mechanisms mediating plaid perception.

Information about the visual angle size of objects is important for maintaining object constancy with variations in viewing distance. Although human observers are quite accurate at judging spatial separations (or cross-sectional size), they are prone to error when there are other spans nearby, as in classical illusions such as the Müller-Lyer illusion. It is possible to reconcile these aspects of size perception by assuming that the size domain is sampled sparsely. It was shown by means of a visual search procedure that the size of objects is processed preattentively and in parallel across the visual field. It was demonstrated that an object's size, rather than its boundary curvature or spatial-frequency content, provides the basis for parallel visual search. It was also shown that texture borders could be substituted for luminance borders, indicating that object boundaries at the relevant spatial scale provide the input to size perception. Parallel processing imposes a severe computational constraint which provides support for the assumption of sparse sampling. An economical model based on several broadly tuned layers of size detectors is proposed to account for the parallel extraction of size, the Weberian behaviour of size discrimination, and the occurrence of strong interference effects in the size domain.

Performance on visual tasks involving the use of motion-defined contours is likely to depend on stimulus strength, but presently there are no empirical or experimental assessments of motion-defined contour strength. Therefore, a matching method was used to estimate the strength of suprathreshold motion-defined edges on a luminance-contrast scale. The perceived strength of a motion-defined contour was expressed as an equivalent luminance contrast; this allowed the use of a single scale which accommodates diverse motion-defined stimuli. Motion-defined edge strength estimated in this manner was an inverted U-shaped function of dot density and dot velocity, and spanned at least a fivefold range of edge strengths. For one observer, maximum motion-defined edge strength was equivalent to 79% luminance contrast, at least thirteen times the contrast detection threshold. The results are interpreted via a simple two-stage model for perceiving motion-defined edges.

It is shown that human observers can use color both for detecting and for discriminating motion. The contributions of chromaticity and luminance to the detection and discrimination of motion are investigated with a high-contrast, nonisoluminant stimulus. The motion stimulus is a rectangular ‘particle’ defined by its luminance and chromaticity, which moves against a background containing luminance noise. Although the luminance noise is found to make achromatic particles undetectable over a large range of luminances, the addition of color to a particle can render it detectable and also enable accurate speed discriminations to be made. The contributions of both luminance and chromaticity were measured. The effect of changing the hue angle of the particle as it moves was also examined, and it was found that the detectability of motion is low in that circumstance.

The approach of an object can be monitored from its optic flow. More specifically, it has been postulated that time-to-collision at constant velocity is perceived by relating visual angle θ to its rate of change θ̇, time-to-collision being θ/θ̇.

This hypothesis is reappraised, and an alternative based on the parameters θ̇ and angular acceleration θ̈ is proposed. The expression 2 θ̇/θ̈ also specifies time-to-collision, with the benefits that it removes reliance on θ and permits time-to-collision to be determined from even momentary perception of an approaching point. This is supported by tests in which subjects responded to computer simulations of approaching objects. A further benefit is that if the object is accelerating rather than at constant velocity, time-to-collision is adjusted by 2 θ̇/θ̈, but not by θ/θ̇.

As time-to-collision increases, however, its cognitive derivation should transfer from optic flow to separate perceptions of distance and speed. It is proposed that when drivers of road vehicles are in potential collision with pedestrians their perception of distance is based primarily on familiar size, resulting in overestimation of size and therefore of time-to-collision with child pedestrians. This is supported by further computer simulations and is corroborated by predicting the effect of that overestimation on certain types of accidents, then testing from national accident statistics.

These analyses indicate that drivers' misperception of time-to-collision has a dramatic effect on the accident rate of child pedestrians. It is proposed that this could be greatly reduced by the provision of remedial measures.

Inspection time (IT) is an index of speed of early visual processing that correlates significantly with measures of higher cognitive ability. An obstacle to the understanding of this important association is the lack of an agreed mechanistic model of the processes involved in IT performance. It is argued here that, whereas in current attempts to model IT processing, stationarity in IT performance is assumed, it is necessary to address this key issue empirically. Two practised subjects undertook 38 400 trials on a standard two-choice visual IT discrimination task. Testing was spread evenly over 60 days. Isotonic regression analysis revealed that both subjects showed nonstationarity in performance over very long periods. The results were not merely due to long-term practice effects; one subject improved in some durations but worsened at others. Therefore, current attempts at modelling in which stationarity in IT performance is assumed must be altered to incorporate the temporal dynamics of IT performance, which might have individual differences.