The role of surface-based representations of shape in visual object recognition

This study contrasted the role of surfaces and volumetric shape primitives in three-dimensional object recognition. Observers (N?=?50) matched subsets of closed contour fragments, surfaces, or volumetric parts to whole novel objects during a whole–part matching task. Three factors were further manipulated: part viewpoint (either same or different between component parts and whole objects), surface occlusion (comparison parts contained either visible surfaces only, or a surface that was fully or partially occluded in the whole object), and target–distractor similarity. Similarity was varied in terms of systematic variation in nonaccidental (NAP) or metric (MP) properties of individual parts. Analysis of sensitivity (d′) showed a whole–part matching advantage for surface-based parts and volumes over closed contour fragments—but no benefit for volumetric parts over surfaces. We also found a performance cost in matching volumetric parts to wholes when the volumes showed surfaces that were occluded in the whole object. The same pattern was found for both same and different viewpoints, and regardless of target–distractor similarity. These findings challenge models in which recognition is mediated by volumetric part-based shape representations. Instead, we argue that the results are consistent with a surface-based model of high-level shape representation for recognition.     

Accessing long-term memory representations during visual change detection

In visual change detection tasks, providing a cue to the change location concurrent with the test image (post-cue) can improve performance, suggesting that, without a cue, not all encoded representations are automatically accessed. Our studies examined the possibility that post-cues can encourage the retrieval of representations stored in long-term memory (LTM). Participants detected changes in images composed of familiar objects. Performance was better when the cue directed attention to the post-change object. Supporting the role of LTM in the cue effect, the effect was similar regardless of whether the cue was presented during the inter-stimulus interval, concurrent with the onset of the test image, or after the onset of the test image. Furthermore, the post-cue effect and LTM performance were similarly influenced by encoding time. These findings demonstrate that monitoring the visual world for changes does not automatically engage LTM retrieval.     

Dissociations of the number and precision of visual short-term memory representations in change detection

The present study dissociated the number (i.e., quantity) and precision (i.e., quality) of visual short-term memory (STM) representations in change detection using receiver operating characteristic (ROC) and experimental manipulations. Across three experiments, participants performed both recognition and recall tests of visual STM using the change-detection task and the continuous color-wheel recall task, respectively. Experiment 1 demonstrated that the estimates of the number and precision of visual STM representations based on the ROC model of change-detection performance were robustly correlated with the corresponding estimates based on the mixture model of continuous-recall performance. Experiments 2 and 3 showed that the experimental manipulation of mnemonic precision using white-noise masking and the experimental manipulation of the number of encoded STM representations using consolidation masking produced selective effects on the corresponding measures of mnemonic precision and the number of encoded STM representations, respectively, in both change-detection and continuous-recall tasks. Altogether, using the individual-differences (Experiment 1) and experimental dissociation (Experiment 2 and 3) approaches, the present study demonstrated the some-or-none nature of visual STM representations across recall and recognition.     

Volatile visual representations: failing to detect changes in recently processed information

Research documenting people’s inability to detect large changes in visual scenes suggests that visual representations may be sparse and volatile, providing no cumulative record of the attended items in a scene. However, these studies have failed to control for attention. Thus, the visual system may construct a cumulative record of all attended stimuli and still miss such changes, because they involve items that were never attended to. In two experiments, subjects saw 12-digit arrays and identified either the highest digit in the array (Experiment 1) or the lowest digit not in the array (Experiment 2). Subsequent change-detection tasks revealed that subjects often failed to detect changes that involved the same digits they had previously identified to perform the digit tasks successfully. This provides additional evidence that our usable visual representations are relatively impoverished and volatile.     

Generation of visual representations

The visual confusability of uppercase letters was manipulated in a successive same-different task to study the conditions under which visual generation from auditory inputs would occur and to investigate the figural specificity of the generated representations. Prior experiments have shown that visual confusions do occur when the initial stimulus is auditory and the second one is visual, which indicates that auditory stimuli can be encoded into visual forms. There has been some suggestion, however, that the generated visual code may have been too abstract to differentiate between the two cases in which letters can appear. In the present experiment, although the confusion effect was not eliminated when the subjects had no advance knowledge regarding the case in which the visual stimulus would appear, the marked confusion effect obtained when the visual stimulus was an uppercase letter was substantially attenuated when the letter appeared in lowercase. This was taken to indicate that the visual characteristics of a generated visual representation may be relatively specific. The results also suggested that subjects may wait until after the second stimulus is presented before they generate the visual representation of the initial auditory stimulus.     

Modelling visual detection: Luminance response non-linearity and internal noise

Two experiments that investigate the effect of various display factors on the detectability of a thin line signal in random visual noise are described. Three statistical decision models are described, together with their ability to account for the results. The first is an “ideal detector” model, the second an “energy integrator” model, and the third a model based upon the operation of retinal ganglion cells which incorporates a gain control mechanism. The ideal detector model fails to give a good account of human performance, whereas the other two models provide a good fit to the data. The digital Laplacian with gain control model has the slight advantage over the energy integrating model in being able to account for a small superiority in the detection of dark as opposed to bright signals. Finally, both models require the inclusion of an estimate of the internal noise of the human visual system to account for the pattern of performance observed under changing conditions of display contrast.     

Neural changes in the ventral and dorsal visual streams during pattern recognition learning

The learning process related to pattern and object recognition is difficult to study because the human brain has a remarkable capacity to recognise complex visual forms from early infancy. In the present study, we investigated on-going neural changes underlying the learning process of visual pattern recognition by means of a device substituting audition for vision. Functional MRI evidenced the gradual pattern recognition-induced recruitment of the ventral visual stream, bilaterally, from learning session 1 to session 3, and a slight decrease in these activation foci from session 3 to session 4. The initial increase in activation is thought to reflect the gradually enhanced visualisation of patterns in the subjects' mind across sessions. By contrast the subsequent decrease reported at the end of the training period is interpreted as the progressive optimisation of neuronal responses elicited by the task. Our results, in accordance with previous observations, suggest that the succession of activation increase and decrease in sensori-motor areas could be a general rule in sensory and sensori-motor learning.     

On the role of outline shape and word-specific visual pattern in the identification of function words: None

A common assumption in the reading literature is that function words are often identified wholly or in part on the basis of outline shape and/or word-specific visual pattern. This assumption was examined in three perceptual identification experiments, a speeded oral reading experiment, and in a lexical decision task where the visual pattern of function words and control items (content words, non-words and control strings) was systematically distorted by cAsE alTeRnA-tIoN. In all five experiments distortion failed to impair performance more on functors than on various control items. This result is argued to be inconsistent with the view that outline shape and word-specific visual pattern typically play a role in the identification of isolated function words.

Shape distortion impaired function words less than controls in perceptual identification, but impaired function and content words equally in speeded naming and lexical decision. These contrasting effects of shape distortion are discussed in terms of a distinction between the uptake of information (lexical access) and phenomenal perception.     

Perceptual completion and object-based representations in short-term visual memory

Object-based representations in visual short-term memory (VSTM) were examined using a change detection memory task. A display comprising two rows of four differently colored elements was followed by a probe display in which only one of the rows reappeared. On same trials, the probed row was identical to the corresponding row in the memory display. On different trials, two of the elements in the probed row had their colors exchanged. In each memory display, a task-irrelevant visual element appeared between the two rows, with the potential to function as an occluder. Performance was enhanced when perceptual completion meant that four, rather than eight, objects were perceived in the memory display and when the probe display revealed that the occluded elements continued behind the occluder. It appears that several forms of representation can co-occur to support VSTM, one of which is object based.     

Implicit learning for probable changes in a visual change detection task

Previous research demonstrates that implicitly learned probability information can guide visual attention. We examined whether the probability of an object changing can be implicitly learned and then used to improve change detection performance. In a series of six experiments, participants completed 120–130 training change detection trials. In four of the experiments the object that changed color was the same shape (trained shape) on every trial. Participants were not explicitly aware of this change probability manipulation and change detection performance was not improved for the trained shape versus untrained shapes. In two of the experiments, the object that changed color was always in the same general location (trained location). Although participants were not explicitly aware of the change probability, implicit knowledge of it did improve change detection performance in the trained location. These results indicate that improved change detection performance through implicitly learned change probability occurs for location but not shape.     

Separate visual representations in the planning and control of action

Evidence for a dichotomy between the planning of an action and its on-line control in humans is reviewed. This evidence suggests that planning and control each serve a specialized purpose utilizing distinct visual representations. Evidence from behavioral studies suggests that planning is influenced by a large array of visual and cognitive information, whereas control is influenced solely by the spatial characteristics of the target, including such things as its size, shape, orientation, and so forth. Evidence from brain imaging and neuropsychology suggests that planning and control are subserved by separate visual centers in the posterior parietal lobes, each constituting part of a larger network for planning and control. Planning appears to rely on phylogenetically newer regions in the inferior parietal lobe, along with the frontal lobes and basal ganglia, whereas control appears to rely on older regions in the superior parietal lobe, along with the cerebellum.     

Prosodic phonological representations early in visual word recognition

Two experiments examined the nature of the phonological representations used during visual word recognition. We tested whether a minimality constraint (R. Frost, 1998) limits the complexity of early representations to a simple string of phonemes. Alternatively, readers might activate elaborated representations that include prosodic syllable information before lexical access. In a modified lexical decision task (Experiment 1), words were preceded by parafoveal previews that were congruent with a target's initial syllable as well as previews that contained 1 letter more or less than the initial syllable. Lexical decision times were faster in the syllable congruent conditions than in the incongruent conditions. In Experiment 2, we recorded brain electrical potentials (electroencephalograms) during single word reading in a masked priming paradigm. The event-related potential waveform elicited in the syllable congruent condition was more positive 250-350 ms posttarget compared with the waveform elicited in the syllable incongruent condition. In combination, these experiments demonstrate that readers process prosodic syllable information early in visual word recognition in English. They offer further evidence that skilled readers routinely activate elaborated, speechlike phonological representations during silent reading.     

Preserved visual representations despite change blindness in infants

Combining theoretical hypotheses of infant cognition and adult perception, we present evidence that infants can maintain visual representations despite their failure to detect a change. Infants under 12 months typically fail to notice a change to an object's height in a covering event. The present experiments demonstrated that 11‐month‐old infants can nevertheless maintain a viable representation of both the pre‐ and post‐change heights despite their ‘change blindness’. These results suggest that infants, like adults, can simultaneously maintain multiple representations, even if they do not optimally use them.     

Symmetry in haptic and in visual shape perception

Four experiments tested the hypothesis that bilateral symmetry is an incidental encoding property in vision, but can also be elicited as an incidental effect in touch, provided that sufficient spatial reference information is available initially for haptic inputs to be organized spatially. Experiment 1 showed that symmetry facilitated processing in vision, even though the task required judgments of stimulus closure rather than the detection of symmetry. The same task and stimuli failed to show symmetry effects in tactual scanning by one finger (Experiment 2). Experiment 3 found facilitating effects for vertically symmetric open stimuli, although not for closed patterns, in two-forefinger exploration when the fore-fingers had previously been aligned to the body midaxis to provide body-centered spatial reference. The one-finger exploration condition again failed to show symmetry effects. Experiment 4 replicated the facilitating effects of symmetry for open symmetric shapes in tactual exploration by the two (previously aligned) forefingers. Closed shapes again showed no effect. Spatial-reference information, finger movements, and stimulus factors in shape perception by touch are discussed.     

Functional representations common to visual perception and imagination

Subjects determined whether probe dots appearing in component squares of a 5 X 5 grid fell on a figure that, depending on the condition, was (a) visually presented as a pattern of darkened squares, (b) only remembered from a preceding presentation of such a pattern, or (c) imaginally generated from a verbal code. The speed and accuracy of the responses to the probes as well as the functional dependencies of the reaction times on structural variables were essentially the same whether the figural pattern was imagined, remembered, or actually seen. Reaction times averaged between 400 and 500 msec and were consistently faster (a) for on-figure than for off-figure responses, (b) for simpler figures that had fewer squares, (c) for off-figure probes that were more distant from the figure, and (d) for on-figure probes that consisted either of more dots on the figure or of a dot at the intersection of a horizontal and a vertical bar. The reults were consistent with a model in which a subject's perceptual or imaginal representations of these forms contain barlike units that respond independently to the probes.     

Superstitious perceptions reveal properties of internal representations

Everyone has seen a human face in a cloud, a pebble, or blots on a wall. Evidence of superstitious perceptions has been documented since classical antiquity, but has received little scientific attention. In the study reported here, we used superstitious perceptions in a new principled method to reveal the properties of unobservable object representations in memory. We stimulated the visual system with unstructured white noise. Observers firmly believed that they perceived the letter S in Experiment 1 and a smile on a face in Experiment 2. Using reverse correlation and computational analyses, we rendered the memory representations underlying these superstitious perceptions.     

Projective invariance and visual shape constancy

As one moves about a table, the projection of its shape on the retina varies enormously, yet the table's shape appears constant. The various retinal images of a single object are nearly congruent in projective geometry. To explain apparent constancy, standard theories of vision assume that the visual system has access to this projective congruence. We present four experiments that undermine this assumption (i.e., the projective thesis). The basic result is that observers' estimates of shape in a simple production task represent gross departures from correct projection, even when observers are given aids to fixation. We manipulate both observer sample and experimental procedure in an attempt to find a source of these persistent errors. Our present hypothesis is that observers lack the sensitivity or implicit knowledge of projective geometry that has been attributed to them.     

The structure of visual pattern associates and pattern goodness

Two tasks were used with a total set of 126 dot patterns. In one task Ss rated the goodness of each pattern. In another task they produced a dot pattern as an associate to each of the patterns used as a stimulus. The distributions of the associates suggest that the total set of patterns is both partitioned and nested. Groups defined by rotation and reflection are partitioned, thus kept intact. These groups in turn form a series of nested subsets., Both partitioning and nesting produce subsets of different size. The size of these subsets is related to pattern goodness, with good patterns coming from small subsets.