Visual pattern recognition: Size preprocessing re-examined

Template theories of visual pattern recognition assume the operation of preprocessing routines to deal with irrelevancies such as discrepancies in stimulus size. In three experiments where size was an irrelevant dimension, observers classified pairs of forms as either “same” or “different”. In Experiment I, the classification “different” was required when the stimuli shared the same form but a different orientation, and “same” when the stimuli shared the same form and orientation. Under these conditions RT was an increasing function of the magnitude of the size disparity between stimuli with equal slopes for “same” and “different” judgements. In Experiment II, “different” classifications were made to stimuli that had different forms, and “same” to figures with the same form. This stimulus set produced a size disparity function that interacted with response type; “different” responses had a shallower slope. Experiment III consisted of a mixed stimulus set drawn from both Experiment I and II. Stimuli that produced additive effects of size disparity and response type in Experiment I now produced an interaction between these two factors similar to the one observed in Experiment II. The results of these experiments are interpreted as evidence that previous contradictory results reported in the literature stem from differences in the way the stimulus set is constructed, and that size transformations can not be a necessary operation, at least when “different” judgements are made. The results are problematic for the view that size disparity effects in matching tasks are easily interpretable in terms of a primitive size normalization stage that precedes any comparison operations.     

Context effects in visual pattern recognition by pigeons

In the experiments described in this paper we examined the effects of contextual stimuli on pigeons’ recognition of visual patterns. Experiment z showed acontext-superiority effect. Specifically, two target forms that were identical except for location in the visual field were not discriminated when presented alone, but the compounds formed when each of these targets was placed between a nearby pair of flanking stimuli were readily discriminated. The size of the context-superiority effect decreased with increasing target-flanker separation. In Experiments 2 and 3 the two targets differed in form rather than spatial location and were readily discriminated in the absence of Hankers. Under these circumstances, adding an identical pair of flankers to each target resulted in a content-inferiority effect; that is, the two target-plus-Hankers compounds were less readily discriminated than the targets atone. The size of the context-inferiority effect decreased with increasing target-flanker separation. The observed effects of context are predictable from the Heinemann-Chase (1990) model of pattern recognition.     

Limited translation invariance of human visual pattern recognition

Visual object recognition is considered to be largely translation invariant. An earlier study (Foster & Kahn, 1985), however, has indicated that recognition of complex novel stimuli is partially specific to location in the visual field: It is significantly easier to determine the identity of two briefly displayed random patterns if both stimuli are presented at the same, rather than at different, locations. In a series ofsame/different discrimination tasks, we characterize the processes underlying this “displacement effect”: Horizontal and vertical translations are equally effective in reducing performance. Making the task more difficult by increasing pattern similarity leads to even higher positional specificity. The displacement effect disappears after rotation or contrast reversal of the patterns, indicating that positional specificity depends on relatively low levels of processing. Control experiments rule out explanations that are independent of visual pattern memory, such as spatial attention, eye movements, or retinal afterimages. Positional specificity of recognition is found only forsame trials. Our results demonstrate that position invariance, a widely acknowledged property of the human visual system, is limited to specific experimental conditions. Normalization models involving mental shifts of an early visual representation or of a window of attention cannot easily account for these findings.     

Size effects in visual recognition memory are determined by perceived size

Recognition memory for shapes has been shown to depend on differences between the size of shapes at the time of encoding and at the time of the memory test (Jolicoeur, 1987). Experiment 1 of the present paper replicates this effect and establishes a set of parameters used in the subsequent experiments. Experiment 2 considers the results of Experiment 1 in light of the distinction between "perceived" size, which, under normal viewing conditions, varies minimally with changes in distance between the observer and object, and "retinal" size, which varies proportionally with viewing distance as an object is moved closer to or farther from an observer. Subjects studied novel shapes and performed a recognition memory test in which the distance from the subject to the viewing screen at the time of testing was different from that at the time of encoding. The viewing distance and the size of the shapes were manipulated such that perceived and retinal sizes were dissociated. The results suggest that the size-congruency effect in memory for visual shape occurs as a result of changes in the perceived size of shapes between the encoding and the testing phases, with little or no contribution of retinal size per se.     

Size scaling in two-dimensional pictorial arrays

The study examined effects of instructional sets (objective, phenomenal, projective, and retinal) on the judged sizes of blocks placed at various “distances” in a pictorial array. Magnitude estimations of size were consistent with previous studies of size constancy in three-dimensional arrays. Chronometric analyses indicated that reaction time increased with distal size, but was not affected by perceived distance. The results suggested that size was scaled relative to a perceptual unit. Instructions affected the nature of the scale unit (proximal vs. distal), but not the scaling process itself.     

Difficulty scaling of recognition items

Usual system: take as scale value the normal deviate corresponding to the areap = (p –p ₀)/(1 –p ₀), wherep is the proportion of correct answers andp ₀ is the chance proportion. Ifp p ₀, the item difficulty is unscalable. Proposed system: take as scale value the normal deviate corresponding to the areap =p ^x , withx so chosen thatp = .5 whenp is half-way betweenp ₀ and 1. Table givesD (=500 + 100z) forp = .00(.01)1.00 and for (= number of alternatives) 2,3,4,5,6,7, .     

Research in visual pattern recognition: the enduring legacy of studies from the 1960s

This article highlights some research in visual pattern perception that was published in the American Journal of Psychology in the 1960s. Although visual perception research has changed substantially since then, and the term visual pattern recognition is no longer in widespread use, the six articles presented here are some of the most influential of those published by the Journal in its long history. The research is described with an emphasis on how a particular publication influenced others and the field in general.     

Size invariance in visual number discrimination

Summary This study deals with the observer's ability to discriminate the numerosity of two random dot-patterns irrespective of their relative size. One of these two patterns was a reference one that was always composed of 32 dots randomly distributed within aK ×K invisible square window (K = 1.92°). The second one was the test pattern with one of the five magnifications (K = 0.64°, 1.28°, 1.92°, 2.56°, 3.20°) and the relative number of dots varied on 11 levels (N = –15, –12, –9, –6, –3, 0, 3, 6, 9, 12, or 15 dots). The observer's task was to indicate which of the two patterns contained more dots. The results show that the stimulus size, as an irrelevant stimulus attribute, can be ignored in the judgements about relative numerosity. This means that the perceived numerosity is size invariant, at least for a 1.6-times magnification and a 3-times reduction of the test pattern. The size invariance observed constrains the range of potential models, since the perceived numerosity can be identified only by means of a feature of the stimulus that will remain invariant after any change in the absolute stimulus size.     

Expertise in tactile pattern recognition

This article explores expertise in tactile object recognition. In one study, participants were trained to differing degrees of accuracy on tactile identification of two-dimensional patterns. Recognition of these patterns, of inverted versions of these patterns, and of subparts of these patterns was then tested. The inversion effect (better recognition of upright than inverted patterns) and the part-whole effect (better recognition of the whole than a part pattern), traditionally considered signatures of visual expertise, were observed for tactile experts but not for novices. In a second study, participants were trained as visual or tactile experts and then tested in the trained and nontrained modalities. Whereas expertise effects were observed in the modality of training, cross-modal transfer was asymmetric; visual experts showed generalization to haptic recognition, but tactile experts did not show generalization to visual recognition. Tactile expertise is not obviously attributable to visual mediation and emerges from domain-general principles that operate independently of modality.     

Size scaling and its effect on letter detection

Two ways of scaling letter size across eccentricity were investigated in a choice reaction time (CRT) task. Experiments 1 and 2 tested cortical magnification theory (M-scaling), while Experiment 3 used scaled sizes drawn from Anstis’s regression formula (A-scaling). Experiment 4 compared both scaling techniques, together with the effect of exposure duration and the absolute size of the foveal letter. Results showed that scaling effectiveness improved when large rather than small foveal letters were used. A-scaling with a large foveal letter provided a good fit for the data at parafoveal locations, but underestimated the letter sizes needed at large eccentricities. M-scaling with a large foveal letter size produced CRTs that were independent of eccentricity. Exposure duration did not substantially affect performance.     

Item performance in visual word recognition

Standard factorial designs in psycholinguistics have been complemented recently by large-scale databases providing empirical constraints at the level of item performance. At the same time, the development of precise computational architectures has led modelers to compare item-level performance with item-level predictions. It has been suggested, however, that item performance includes a large amount of undesirable error variance that should be quantified to determine the amount of reproducible variance that models should account for. In the present study, we provide a simple and tractable statistical analysis of this issue. We also report practical solutions for estimating the amount of reproducible variance for any database that conforms to the additive decomposition of the variance. A new empirical database consisting of the word identification times of 140 participants on 120 words is then used to test these practical solutions. Finally, we show that increases in the amount of reproducible variance are accompanied by the detection of new sources of variance.     

Perceptual grouping in visual word recognition

Four experiments are presented in which printed texts are read for their meaning. Some of the texts were mutilated by altering the size of selected letters. In Experiments 1, 2, and 3, the number of words mutilated per passage and the number of letters changed per word were both manipulated. In all three experiments, reading was slowed as a function of the number of words changed per passage, while the number of letters changed per word had a much smaller effect. The interaction between the number of words and number of letters changed was not significant in any of the experiments. It is difficult to explain these results merely in terms of changes in the discriminability of letters. In Experiment 2 all uppercase text was used, which argues against an explanation in terms of supraletter features such as word envelope. We propose an explanation in terms of visual attention and the perceptual grouping required prior to feature recognition. The last experiment supports this explanation through the counterintuitive finding that adding letters of intermediate size can improve legibility by allowing grouping processes to associate large and small letters as belonging to the same word object.     

Psychoticism and visual recognition thresholds

This paper presents data which are of relevance in assessing the validity of the P scale of the Eysenck Personality Questionnaire (EPQ; Eysenck and Eysenck. 1975). Frith (1979) suggested that a defect in inhibitory selective mechanisms constitutes the cognitive disorder underlying the positive symptomatology of schizophrenia: hallucinations, delusions and thought disorder. A priming paradigm was used to measure the operation of inhibitory mechanisms. It was predicted that, if psychosis is conceptualized as a pathological exaggeration of the features characterizing the trait of psychoticism then high P scorers should show a reduced inhibition effect. This prediction was confirmed, a negative correlation was found between P and the size of the inhibition effect. No correlation with IQ or other personality variables from the EPQ was found.