Stereo disparity facilitates view generalization during shape recognition for solid multipart objects

Current theories of object recognition in human vision make different predictions about whether the recognition of complex, multipart objects should be influenced by shape information about surface depth orientation and curvature derived from stereo disparity. We examined this issue in five experiments using a recognition memory paradigm in which observers (N = 134) memorized and then discriminated sets of 3D novel objects at trained and untrained viewpoints under either mono or stereo viewing conditions. In order to explore the conditions under which stereo-defined shape information contributes to object recognition we systematically varied the difficulty of view generalization by increasing the angular disparity between trained and untrained views. In one series of experiments, objects were presented from either previously trained views or untrained views rotated (15°, 30°, or 60°) along the same plane. In separate experiments we examined whether view generalization effects interacted with the vertical or horizontal plane of object rotation across 40° viewpoint changes. The results showed robust viewpoint-dependent performance costs: Observers were more efficient in recognizing learned objects from trained than from untrained views, and recognition was worse for extrapolated than for interpolated untrained views. We also found that performance was enhanced by stereo viewing but only at larger angular disparities between trained and untrained views. These findings show that object recognition is not based solely on 2D image information but that it can be facilitated by shape information derived from stereo disparity.     

Flipping and spinning: Spatial transformation procedures in the identification of rotated natural objects

The proposal that identification of inverted objects is accomplished by either a relatively slow rotation in the picture plane or a faster rotation in the depth plane about the horizontal axis was tested. In Experiment 1, subjects decided whether objects at 0° or 180° corresponded to previously learned normal views of the upright objects, or were mirror images. Instructions to mentally flip an inverted object in the depth plane to the upright produced faster decision times than did instructions to mentally spin the object in the picture plane. In Experiment 2, the effects of orientation were compared across an object-naming task and a normal-mirror task for six orientations from 0° to 300°. In the normal-mirror task, objects at 180° were cued for rotation in the picture plane or in the depth plane in equal numbers. The naming function for one group of subjects did not differ from the normalmirror function where inverted objects had been mentally rotated to the upright. For both functions, response time (RT) increased linearly from 0° to 180° and the slopes did not differ. The naming function for a second group of subjects did not differ from the normal-mirror function where inverted objects had been mentally flipped to the upright. For both functions, RT increased linearly at a similar rate from 0° to 120°, but decreased from 120° to 180°. The results are discussed in terms of theories of orientation-specific identification.     

Recognizing rotated views of objects: Interpolation versus generalization by humans and pigeons

Pigeons and humans were trained to discriminate between pictures of three-dimensional objects that differed in global shape. Each pair of objects was shown at two orientations that differed by a depth rotation of 90° during training. Pictures of the objects at novel depth rotations were then tested for recognition. The novel test rotations were 30°, 45°, and 90° from the nearest trained orientation and were either interpolated between the trained orientations or extrapolated outside of the training range. For both pigeons and humans, recognition accuracy and/or speed decreased as a function of distance from the nearest trained orientation. However, humans, but not pigeons, were more accurate in recognizing novel interpolated views than novel extrapolated views. The results suggest that pigeons’ recognition was based on independent generalization from each training view, whereas humans showed view-combination processes that resulted in a benefit for novel views interpolated between the training views.     

Measuring the speed of newborn object recognition in controlled visual worlds

How long does it take for a newborn to recognize an object? Adults can recognize objects rapidly, but measuring object recognition speed in newborns has not previously been possible. Here we introduce an automated controlled‐rearing method for measuring the speed of newborn object recognition in controlled visual worlds. We raised newborn chicks (Gallus gallus) in strictly controlled environments that contained no objects other than a single virtual object, and then measured the speed at which the chicks could recognize that object from familiar and novel viewpoints. The chicks were able to recognize the object rapidly, at presentation rates of 125 ms per image. Further, recognition speed was equally fast whether the object was presented from familiar viewpoints or novel viewpoints (30° and 60° azimuth rotations). Thus, newborn chicks can recognize objects across novel viewpoints within a fraction of a second. These results demonstrate that newborns are capable of both rapid and invariant object recognition at the onset of vision.     

Determining the orientation of depth-rotated familiar objects

How does the human visual system determine the depth-orientation of familiar objects? We examined reaction times and errors in the detection of 15° differences in the depth orientations of two simultaneously presented familiar objects, which were the same objects (Experiment 1) or different objects (Experiment 2). Detection of orientation differences was best for 0° (front) and 180° (back), while 45° and 135° yielded poorer results, and 90° (side) showed intermediate results, suggesting that the visual system is tuned for front, side and back orientations. We further found that those advantages are due to orientation-specific features such as horizontal linear contours and symmetry, since the 90° advantage was absent for objects with curvilinear contours, and asymmetric object diminished the 0° and 180° advantages. We conclude that the efficiency of visually determining object orientation is highly orientation-dependent, and object orientation may be perceived in favor of front-back axes.     

Spatial updating during locomotion does not eliminate viewpoint-dependent visual object processing

Two experiments were conducted to investigate whether locomotion to a novel test view would eliminate viewpoint costs in visual object processing. Participants performed a sequential matching task for object identity or object handedness, using novel 3-D objects displayed in a head-mounted display. To change the test view of the object, the orientation of the object in 3-D space and the test position of the observer were manipulated independently. Participants were more accurate when the test view was the same as the learned view than when the views were different no matter whether the view change of the object was 50° or 90°. With 50° rotations, participants were more accurate at novel test views caused by participants’ locomotion (object stationary) than caused by object rotation (observer stationary) but this difference disappeared when the view change was 90°. These results indicate that facilitation of spatial updating during locomotion occurs within a limited range of viewpoints, but that such facilitation does not eliminate viewpoint costs in visual object processing.     

Achieving visual object constancy across plane rotation and depth rotation.

Visual object constancy is the ability to recognise an object from its image despite variation in the image when the object is viewed from different angles. I describe research which probes the human visual system's ability to achieve object constancy across plane rotation and depth rotation. I focus on the ecologically important case of recognising familiar objects, although the recognition of novel objects is also discussed. Cognitive neuropsychological studies of patients with specific deficits in achieving object constancy are reviewed, in addition to studies which test neurally intact subjects. In certain cases, the recognition of invariant features allows objects to be recognised irrespective of the view depicted, particularly if small, distinctive sets of objects are presented repeatedly. In contrast, in most situations, recognition is sensitive to both the view in-plane and in-depth from which an object is depicted. This result suggests that multiple, view-specific, stored representations of familiar objects are accessed in everyday, entry-level visual recognition, or that transformations such as mental rotation or interpolation are used to transform between retinal images of objects and view-specific, stored representations.     

Dissociating viewpoint costs in mental rotation and object recognition

In a mental rotation task, participants must determine whether two stimuli match when one undergoes a rotation in 3-D space relative to the other. The key evidence for mental rotation is the finding of a linear increase in response times as objects are rotated farther apart. This signature increase in response times is also found in recognition of rotated objects, which has led many theorists to postulate mental rotation as a key transformational procedure in object recognition. We compared mental rotation and object recognition in tasks that used the same stimuli and presentation conditions and found that, whereas mental rotation costs increased relatively linearly with rotation, object recognition costs increased only over small rotations. Taken in conjunction with a recent brain imaging study, this dissociation in behavioral performance suggests that object recognition is based on matching of image features rather than on 3-D mental transformations.     

Object permanence in dogs: Invisible displacement in a rotation task

Dogs were tested for object permanence using an invisible displacement in which an object was hidden in one of two containers at either end of a beam and the beam was rotated. Consistent with earlier research, when the beam was rotated 180°, the dogs failed to find the object. However, when the beam was rotated only 90°, they were successful. Furthermore, when the dogs were led either 90° or 180° around the apparatus, they were also successful. In a control condition, when the dogs could not see the direction of the 90° rotation, they failed to find the object. The results suggest that the 180° rotation may produce an interfering context that can be reduced by rotating the apparatus only 90° or by changing the dogs’ perspective. Once the conflict is eliminated, dogs show evidence of object permanence that includes invisibly displaced objects.     

Haptic object recognition: how important are depth cues and plane orientation?

Raised-line drawings of familiar objects are very difficult to identify with active touch only. In contrast, haptically explored real 3-D objects are usually recognised efficiently, albeit slower and less accurately than with vision. Real 3-D objects have more depth information than outline drawings, but also extra information about identity (eg texture, hardness, temperature). Previous studies have not manipulated the availability of depth information in haptic object recognition whilst controlling for other information sources, so the importance of depth cues has not been assessed. In the present experiments, people named plastic small-scale models of familiar objects. Five versions of bilaterally symmetrical objects were produced. Versions varied only in the amount of depth information: minimal for cookie-cutter and filled-in outlines, partial for squashed and half objects, and full for 3-D models. Recognition was faster and much more accurate when more depth information was available, whether exploration was with both hands or just one finger. Novices found it almost impossible to recognise objects explored with two hand-held probes whereas experts succeeded using probes regardless of the amount of depth information. Surprisingly, plane misorientation did not impair recognition. Unlike with vision, depth information, but not object orientation, is extremely important for haptic object recognition.     

Quantifying the role of context in visual object recognition

An object's context may serve as a source of information for recognition when the object's image is degraded. The current study aimed to quantify this source of information. Stimuli were photographs of objects divided into quantized blocks. Participants decreased block size (increasing resolution) until identification. Critical resolution was compared across three conditions: (1) when the picture of the target object was shown in isolation, (2) in the object's contextual setting where that context was unfamiliar to the participant, and (3) where that context was familiar to the participant. A second experiment assessed the role of object familiarity without context. Results showed a profound effect of context: Participants identified objects in familiar contexts with minimal resolution. Unfamiliar contexts required higher-resolution images, but much less so than those without context. Experiment 2 found a much smaller effect of familiarity without context, suggesting that recognition in familiar contexts is primarily based on object-location memory.     

Seeing two sides at once: effects of viewpoint and object structure on recognizing three-dimensional objects

Five experiments demonstrated that adults can identify certain novel views of 3-dimensional model objects on the basis of knowledge of a single perspective. Geometrically irregular contour (wire) and surface (clay) objects and geometrically regular surface (pipe) objects were accurately recognized when rotated 180 degrees about the vertical (y) axis. However, recognition accuracy was poor for all types of objects when rotated around the y-axis by 90 degrees. Likewise, more subtle rotations in depth (i.e., 30 degrees and 60 degrees) induced decreases in recognition of both contour and surface objects. These results suggest that accurate recognition of objects rotated in depth by 180 degrees may be achieved through use of information in objects' 2-dimensional bounding contours, the shapes of which remain invariant over flips in depth. Consistent with this interpretation, a final study showed that even slight rotations away from 180 degrees cause precipitous drops in recognition accuracy.     

Rotating objects to determine orientation, not identity: evidence from a backward-masking/dual-task procedure

The effects of picture-plane rotations on times taken to name familiar objects (RTs) may reflect a process of mental rotation to stored viewpoint-specific representations: the rotate-to-recognize hypothesis. Alternatively, mental rotation might be used after stored object representations are activated by a misoriented stimulus in order to verify a weak or distorted shape percept: the double-checking hypothesis. We tested these two accounts of rotation effects in object recognition by having subjects verify the orientations (to within 90 degrees) and basic-level names of 14-msec, backward-masked depictions of common objects. The stimulus-mask interval (SOA) varied from 14 to 41 msec, permitting interpolation of the SOA required for 75% accuracy (SOAc). Whereas the SOAc to verify orientation increased with rotation up to 180 degrees, the SOAc to verify identity was briefer and asymptoted at approximately 60 degrees. We therefore reject the rotate-to-recognize hypothesis, which implies that SOAc should increase steadily with rotation in both tasks. Instead, we suggest that upright and near-upright stimuli are matched by a fast direct process and that misoriented stimuli are matched at a featural level by a slightly slower view-independent process. We also suggest that rotation effects on RTs reflect a postrecognition stage of orientation verification: the rotate-to-orient hypothesis, a version of double-checking that also explains the well-known reduction in orientation effects on RTs when naming repeated objects.     

Recognizing familiar objects by hand and foot: Haptic shape perception generalizes to inputs from unusual locations and untrained body parts

The limits of generalization of our 3-D shape recognition system to identifying objects by touch was investigated by testing exploration at unusual locations and using untrained effectors. In Experiments 1 and 2, people found identification by hand of real objects, plastic 3-D models of objects, and raised line drawings placed in front of themselves no easier than when exploration was behind their back. Experiment 3 compared one-handed, two-handed, one-footed, and two-footed haptic object recognition of familiar objects. Recognition by foot was slower (7 vs. 13 s) and much less accurate (9 % vs. 47 % errors) than recognition by either one or both hands. Nevertheless, item difficulty was similar across hand and foot exploration, and there was a strong correlation between an individual’s hand and foot performance. Furthermore, foot recognition was better with the largest 20 of the 80 items (32 % errors), suggesting that physical limitations hampered exploration by foot. Thus, object recognition by hand generalized efficiently across the spatial location of stimuli, while object recognition by foot seemed surprisingly good given that no prior training was provided. Active touch (haptics) thus efficiently extracts 3-D shape information and accesses stored representations of familiar objects from novel modes of input.     

Event-related potential (ERP) indices of infants' recognition of familiar and unfamiliar objects in two and three dimensions

We measured infants' recognition of familiar and unfamiliar 3-D objects and their 2-D representations using event-related potentials (ERPs). Infants differentiated familiar from unfamiliar objects when viewing them in both two and three dimensions. However, differentiation between the familiar and novel objects occurred more quickly when infants viewed the object in 3-D than when they viewed 2-D representations. The results are discussed with respect to infants' recognition abilities and their understanding of real objects and representations. This is the first study using 3-D objects in conjunction with ERPs in infants, and it introduces an interesting new methodology for assessing infants' electrophysiological responses to real objects.     

Object pose: perceiving 3-d shape as sticks and slabs

Estimating the pose (three-dimensional orientation) of objects is an important aspect of 3-D shape perception. We studied the ability of observers to match the pose of the principal axes of an object with the pose of a cross consisting of three perpendicular axes. For objects, we used a long and a flat spheroid and eight symmetric objects with aspect ratios of dimensions of approximately 4:2:1. Stimulus cues were the contour and stereo for the spheroids, and contour, stereo, and shading for the symmetric objects. In addition, the spheroids were shown with or without surface texture and with or without active motion. Results show that observers can perform the task with standard deviations of a few degrees, though biases could be as large as 30 degrees. The results can be naturally decomposed in viewer-centered coordinates, and it turns out that the estimation of orientation in the frontoparallel plane (tilt) is more precise than estimation of orientation in depth (slant, roll). A comparison of long and flat spheroids shows that sticks lead to better performance than do slabs. This can even be the case within the same object; the pose of the stick-like aspect is seen with more precision than is the pose of the slab-like aspect. The largest biases occurred when the spheroids were displayed with the binocular contour as the only cue. We can explain these biases by assuming that subjects' settings are influenced by the orientation of the rim.     

The effects of plane rotation on the recognition of brief masked pictures of familiar objects

Two experiments examined the effects of plane rotation on the recognition of briefly displayed pictures of familiar objects, using a picture—word verification task. Mirroring the results of earlier picture naming studies (Jolicoeur, 1985; Jolicoeur & Milliken, 1989), plane rotation away from a canonical upright orientation reduced the efficiency of recognition, although in contrast to the results from picture naming studies, the rotation effects were not reduced with experience with the stimuli. However, the rotation effects were influenced by the visual similarity of the distractor objects to the picture of the object presented, with greater orientation sensitivity being observed when visually similar distractors were presented. We suggest that subjects use orientation-sensitive representations to recognize objects in both the present unspeeded verification and in the earlier speeded naming tests of picture identification.     

视觉表象操作加工的眼动实验研究

本研究通过视觉表象旋转和扫描的眼动实验探讨表象的心理表征方式。实验一结果表明,眼动指标具有与反应时相类似的旋转角度效应。实验二结果显示,表象扫描的反应时和眼动指标都具有与知觉扫描加工一样的距离效应。由此可以认为,表象眼动与知觉眼动模式具有相似性;表象具有相对独立的心理表征方式并有其特殊的加工过程;表象的心理表征可以是形象表征,而非一定是抽象的命题或符号表征     

Invariant recognition of natural objects in the presence of shadows

Shadows are frequently present when we recognize natural objects, but it is unclear whether they help or hinder recognition. Shadows could improve recognition by providing information about illumination and 3-D surface shape, or impair recognition by introducing spurious contours that are confused with object boundaries. In three experiments, we explored the effect of shadows on recognition of natural objects. The stimuli were digitized photographs of fruits and vegetables displayed with or without shadows. In experiment 1, we evaluated the effects of shadows, color, and image resolution on naming latency and accuracy. Performance was not affected by the presence of shadows, even for gray-scale, blurry images, where shadows are difficult to identify. In experiment 2, we explored recognition of two-tone images of the same objects. In these images, shadow edges are difficult to distinguish from object and surface edges because all edges are defined by a luminance boundary. Shadows impaired performance, but only in the early trials. In experiment 3, we examined whether shadows have a stronger impact when exposure time is limited, allowing little time for processing shadows; no effect of shadows was found. These studies show that recognition of natural objects is highly invariant to the complex luminance patterns caused by shadows.