Detection of glass patterns by pigeons and humans: implications for differences in higher-level processing.

Glass patterns have been used to examine mechanisms underlying form perception. The current investigation compared detection of Glass patterns by pigeons and humans and provides evidence for substantial species differences in global form perception. Subjects were required to discriminate, on a simultaneous display, a random dot pattern from a Glass pattern. Four different randomly presented Glass patterns were used (concentric, radial, parallel-vertical, and parallel-horizontal). Detection thresholds were measured by degrading the Glass patterns through the addition of random noise. For both humans and pigeons, discrimination decreased systematically with the addition of noise. Humans showed detection differences among the four patterns, with lowest thresholds to radial and concentric patterns and highest thresholds to the parallel-horiZontal pattern. Pigeons did not show a detection difference across the four patterns. Implications for differences in neural processing of complex forms are discussed.

Learning to recognize objects.

Several aspects of systems for learning pattern or object recognition rules are discussed. First, how are recognition rules developed and to what extent is structural pattern information embedded into these recognition rules. Second, how are these rules applied to the recognition of complex patterns such as objects embedded in scenes and how is evidence from different rules combined into a single evidence vector. Third, how can learned rules be improved through performance evaluation and feedback to rule generation stages.

Perception of coherent motion in random dot displays by pigeons and humans.

Pigeons and humans were required to discriminate coherent from random motion in dynamic random dot displays. Coherence and velocity thresholds were determined for both species, and both thresholds were found to be substantially higher for pigeons than for humans. The results are discussed with reference to differences in motion processing in mammals and birds. It is suggested that the inferior motion sensitivity of pigeons can be attributed to poorer spatiotemporal motion integration.

The attentional blink with targets in different spatial locations.

When two targets (T1 and T2) are displayed in rapid succession, accuracy of T2 identification varies as a function of the temporal lag between the targets (attentional blink, AB). In some studies, performance has been found to be most impaired at Lag 1--namely, when T2 followed T1 directly. In other studies, T2 performance at Lag 1 has been virtually unimpaired (Lag 1 sparing). In the present work, we examined how Lag 1 sparing is affected by attentional switches between targets displayed in the same location or in different locations. We found that Lag 1 sparing does not occur when a spatial shift is required between T1 and T2. This suggests that attention cannot be switched to a new location while the system is busy processing another stimulus. The results are explained by a modified version of an attentional gating model (Chun & Potter, 1995; Shapiro & Raymond, 1994).

Common reference frame for neural coding of translational and rotational optic flow.

Self-movement of an organism through the environment is guided jointly by information provided by the vestibular system and by visual pathways that are specialized for detecting 'optic flow'. Motion of any object through space, including the self-motion of organisms, can be described with reference to six degrees of freedom: rotation about three orthogonal axes, and translation along these axes. Here we describe neurons in the pigeon brain that respond best to optic flow resulting from translation along one of the three orthogonal axes. We show that these translational optic flow neurons, like rotational optic flow neurons, share a common spatial frame of reference with the semicircular canals of the vestibular system. The three axes to which these neurons respond best are the vertical axis and two horizontal axes orientated at 45 degrees to either side of the body midline.

Bootstrap estimates of the statistical accuracy of thresholds obtained from psychometric functions.

Estimates of the accuracy of a threshold obtained from a psychometric function are often based on asymptotic theory. When the number of trails is small, however, these estimates may be untrustworthy. A computer program is described that uses a move reliable bootstrap approach to obtaining estimates of the standard deviation and confidence limits of a threshold and of the slope and spread of the psychometric function for any criterion level of performance.

Visual learning of patterns and objects.

We discuss automatic rule generation techniques for learning relational properties of 2D visual patterns and 3D objects from training samples where the observed feature values are continuous. In particular, we explore a conditional rule generation method that defines patterns (or objects) in terms of ordered lists of bounds on unary (pattern part) and binary (part relation) features. The technique, termed conditional rule generation, was developed to integrate relational structure representations of patterns and the generalization characteristics of evidenced-based systems. We show how this technique can be used for recognition of complex patterns and of objects in scenes. Further, we show the extent to which the learned rules can identify patterns and objects that have undergone nonrigid distortions.

Machine learning paradigms for pattern recognition and image understanding.

In this paper some issues are considered related to the encoding of spatial information and associated perceptual learning algorithms which, it is claimed, are necessary for robust pattern and object recognition in multi-object (natural) scenes. The types of learning requirements within a 'recognition-by-parts' paradigm are contrasted with findings from alternative models.

Psychophysical evidence of a sustained input to directionally selective motion mechanisms.

Human psychophysical evidence congruent with neurophysiological findings of a sustained input to directionally selective motion sensors in cat visual cortex is reported. Apparent motion was produced by displaying a group of dots in two frames (F1 and F2), where F2 was a translated version of F1. All stimulus sequences included a period during which F1 and F2 were displayed concurrently (combined images) and a period during which only F1 or F2 was on display (single images). There were three stimulus sequences: a display beginning with combined and ending with single image, a display beginning with single and ending with combined image, and a display beginning with F1, continuing with combined image, and ending with F2. Six durations of single and of combined images (10, 20, 40, 80, 160, and 320 ms) were crossed factorially in each stimulus sequence. Directional motion was seen easily at long durations of the single image in all stimulus sequences, as would be expected on the basis of a sustained input to the directional-motion-sensing mechanisms. Perception of directional motion improved with the duration of single images, but declined as the duration of combined images was increased. Baker and Cynader's model could account for the effect of duration of single images, but not for the effect of duration of combined images. An elaborated version of the model provides a good qualitative match to all empirical findings.

Inverse-intensity effect in duration of visible persistence.

Duration of visible persistence can vary inversely with stimulus intensity. This inverse-intensity effect is obtained by varying the intensity of the stimuli or of the background, provided that the variations extend into the mesopic range. A similar relationship--known as the Ferry-Porter law--holds for the critical frequency at fusion (CFF). The authors propose that studies of CFF, 2-pulse threshold, and visible persistence can be encompassed within 1 conceptual framework in which the effect is modeled by the progressive reduction in the temporal extent of the positive phase of the system's response as the level of light adaptation changes from scotopic to photopic. In this context, the authors present an integrative scheme in which G. Sperling and M. M. Sondhi's (1968) formal model and M. Coltheart's (1980) neurophysiological conjecture are shown to be compatible and complementary accounts of the effect.

Motion and metacontrast with simultaneous onset of stimuli.

Coherent directional motion can be seen if an image is displayed in two sequential frames (F1 and F2), where F2 is a translated version of F1. A similar two-frame sequence can produce metacontrast masking: the visibility of a leading target (F1) is reduced by a trailing, spatially nonoverlapping mask (F2). Strict temporal succession of the stimuli has been considered essential for both motion and masking. This requirement for a minimum stimulus-onset asynchrony (SOA) is known as the SOA law. Contrary to the SOA law, we found that motion and masking can be obtained with simultaneous onsets of the stimuli, provided that F2 outlasts F1. We compared motion and metacontrast with simultaneous onsets of the stimuli (SIM paradigm) with the traditional paradigm in which an interstimulus interval (ISI) is inserted between the leading and the trailing stimuli (ISI paradigm). We studied the effects in light-adapted and in dark-adapted viewing, each over a wide range of stimulus intensities. Homologous results were obtained with the two paradigms, thus disconfirming the SOA law. Models of motion sensors, such as that proposed by Reichardt [in Sensory Communication, W. A. Rosenblith, ed. (MIT Press, Cambridge, Mass., 1961), p. 303], are inherently capable of explaining the motion results obtained with both paradigms. The masking results with the SIM paradigm disconfirm theories based on onset-locked slow excitatory and fast inhibitory responses but can be explained in terms of Bridgeman's network model [Bull. Math. Biol. 40, 605 (1978)]. In light of the results obtained with the two paradigms, we discuss, and tentatively support, the suggestion that motion and metacontrast may be complementary parts of a unitary perceptual system.

Computational approaches to human pattern recognition.

This paper consolidates recent findings on how humans detect and recognize patterns and considers computational procedures which reflect observed performance. A multi-level correlation model for spatial information processing is proposed and used to interpret past results on human psychophysical performance.

On the confounding effects of phosphor persistence in oscilloscopic displays.

Phosphor persistence has been a source of confounding in studies of temporal integration in vision. We examined the confounding by assessing the effects of the persistence of two commonly-used phosphors (P15 and P31) on performance of a temporal-integration task. In one experiment we eliminated the visibility of phosphor persistence by closing two mechanical shutters upon display termination. In a second experiment we estimated the duration of phosphor persistence by displaying the image behind closed shutters which opened upon display termination. No detectable persistence was every produced by P15 phosphor. By contrast, P31 phosphor produced persistence that lasted several hundred milliseconds even when a veiling light was projected on the screen. We ascribe the earlier instances of confounding to inadequate interpretation of the technical data on phosphor decay.

Automated detection and classification of breast tumors.

This paper presents a new method for the mammographic detection and classification of two types of breast tumors, stellate lesions and circumscribed lesions. The method assumes that both types of tumors appear as approximately circular, bright masses with a fuzzy boundary and that stellate lesions are in addition surrounded by a radiating structure of sharp, fine lines. Experimental results for a set of 27 mammograms are presented and the method is shown to have a high detection rate and an extremely low false positive rate.

Automated detection of breast tumors using the asymmetry approach.

A method for automated detection of breast tumors in mammograms is presented. The method uses the asymmetry principle: Strong structural asymmetries between corresponding regions in the left and right breast are taken as evidence for the possible presence of a tumor in that region. Asymmetry detection is achieved in two steps. First, mammograms are aligned, compensating for possible differences in size and shape between the two breasts. Second, asymmetry between corresponding positions is determined using a combination of several asymmetry measures, each responding to different types of asymmetries. Results obtained with a set of mammograms indicate that this method can improve the sensitivity and reliability of systems for automated detection of breast tumors.

Displacement limit (dmax) of sampled directional motion: direct and indirect estimates.

The maximum displacement at which directional motion can be seen, known as dmax, has been said to define the spatial limits of the short-range motion system. Turano and Pantle (1985) used duration of motion aftereffect (MAE) to estimate the spatial limit of the short-range system, the assumption that dmax (a direct measure of motion perception) and MAE (an indirect measure) are equivalent indices of the same underlying perceptual process. In a series of four experiments, we examined this assumption by measuring dmax and duration of MAE across a range of displacements, stimulus waveforms (sine- or square-wave gratings), and spatial frequencies. We found that dmax and duration of MAE were affected differently by changes in the same variables. Therefore, we concluded that the two indices cannot be regarded as equivalent measures of the spatial limits of the short-range process. Two novel effects that separated MAE from motion detection are described, and suggestions for exploring them are outlined.

On the half-cycle displacement limit of sampled directional motion.

We employed filtered random-dot kinematograms to determine the maximum displacement (dmax) at which sampled directional motion was reliably detected. The images were produced using ideal band-pass filters that varied in lower cut-off frequency (f1), in bandwidth, and in range (alpha) of component orientations that were passed. Results showed that dmax, expressed in cycles of f1, increased with f1 and alpha, and decreased with bandwidth. In many conditions, dmax exceeded half a cycle of f1, a result that appears to contradict predictions from quadrature models of motion detection. However, an account that does not violate the half-cycle limit can be given on two assumptions. First, motion perception is mediated by a population of orientation and frequency-selective sensors that respond correctly to displacements up to half a cycle in the preferred direction. Second, the outputs from all sensors (notably including off-axis sensors) are linearly summated to yield perception of motion. A computer simulation based on these assumptions provided a remarkably close fit to the psychophysical data.

Perception of directional sampled motion in relation to displacement and spatial frequency: evidence for a unitary motion system.

Perception of directional motion was studied by displaying two images (F1 and F2) in rapid succession. The two images were identical except for a horizontal displacement of F2 with respect to F1. Observers reported the direction of horizontal motion over a wide range of displacements. The stimuli in Experiment 1 were one-dimensional gratings with spatial frequency between 0.125 and 6 c/deg. Motion was seen at all displacements to almost 0.5 cycles (counterphase) and remained invariant across spatial frequencies. In Experiment 2 the stimuli were band-pass filtered random-dot patterns. The bandwidth of the filters was 1 octave, and centre frequencies ranged from 0.75 to 12 c/deg. In every case, the response functions exhibited quasi-periodic oscillations related to structural properties of the images. One-dimensional analyses based on autocorrelation did not provide a satisfactory account of the data. By contrast, the data were fitted successfully by a two-dimensional analysis that integrated the responses of neighbouring motion detectors so as to yield a smooth motion flow field from which left-right directional motion could be derived. Practically and conceptually, the outcome supports a unitary motion system as distinct from separate systems subserving short-range and long-range motion.

A model of visible persistence and temporal integration.

A quantitative model of temporal integration and visible persistence is described and tested. The model treats visible persistence as resulting from processing activity within sustained visual channels whose temporal response is modelled using a second-order control system. Temporal integration of two successive stimuli is assumed to be enabled by the overlap between the two periods of activity and to be impaired by the non-overlapping activity. The model successfully predicts the effects of inter-stimulus interval and of stimulus duration on goodness of temporal integration.