Effect of image orientation and size on object recognition: responses of single units in the macaque monkey temporal cortex

This study examined how cells in the temporal cortex code orientation and size of a complex object. The study focused on cells selectively responsive to the sight of the head and body but unresponsive to control stimuli. The majority of cells tested (19/26, 73%) were selectively responsive to a particular orientation in the picture plane of the static whole body stimulus, 7/26 cells showed generalisation responding to all orientations (three cells with orientation tuning superimposed on a generalised response). Of all cells sensitive to orientation, the majority (15/22, 68%) were tuned to the upright image. The majority of cells tested (81%, 13/16) were selective for stimulus size. The remaining cells (3/16) showed generalisation across four-fold decrease in size from life-sized. All size-sensitive cells were tuned to life-sized stimuli with decreasing responses to stimuli reduced from life-size. These results do not support previous suggestions that cells responsive to the head and body are selective to view but generalise across orientation and size. Here, extensive selectivity for size and orientation is reported. It is suggested that object orientation and size-specific responses might be pooled to obtain cell responses that generalise across size and orientation. The results suggest that experience affects neuronal coding of objects in that cells become tuned to views, orientation, and image sizes that are commonly experienced. Models of object recognition are discussed.     

EFFECT OF IMAGE ORIENTATION AND SIZE ON OBJECT RECOGNITION: RESPONSES OF SINGLE UNITS IN THE MACAQUE MONKEY TEMPORAL CORTEX

This study examined how cells in the temporal cortex code orientation and size of a complex object. The study focused on cells selectively responsive to the sight of the head and body but unresponsive to control stimuli. The majority of cells tested (19/26, 73%) were selectively responsive to a particular orientation in the picture plane of the static whole body stimulus, 7/26 cells showed generalisation responding to all orientations (three cells with orientation tuning superimposed on a generalised response). Of all cells sensitive to orientation, the majority (15/22, 68%) were tuned to the upright image. The majority of cells tested (81%, 13/16) were selective for stimulus size. The remaining cells (3/16) showed generalisation across four-fold decrease in size from life-sized. All size-sensitive cells were tuned to life-sized stimuli with decreasing responses to stimuli reduced from life-size. These results do not support previous suggestions that cells responsive to the head and body are selective to view but generalise across orientation and size. Here, extensive selectivity for size and orientation is reported. It is suggested that object orientation and size-specific responses might be pooled to obtain cell responses that generalise across size and orientation. The results suggest that experience affects neuronal coding of objects in that cells become tuned to views, orientation, and image sizes that are commonly experienced. Models of object recognition are discussed.     

Similarity dependency of the change in ERP component N1 accompanying with the object recognition learning

Performance during object recognition across views is largely dependent on inter-object similarity. The present study was designed to investigate the similarity dependency of object recognition learning on the changes in ERP component N1. Human subjects were asked to train themselves to recognize novel objects with different inter-object similarity by performing object recognition tasks. During the tasks, images of an object had to be discriminated from the images of other objects irrespective of the viewpoint. When objects had a high inter-object similarity, the ERP component, N1 exhibited a significant increase in both the amplitude and the latency variation across objects during the object recognition learning process, and the N1 amplitude and latency variation across the views of the same objects decreased significantly. In contrast, no significant changes were found during the learning process when using objects with low inter-object similarity. The present findings demonstrate that the changes in the variation of N1 that accompany the object recognition learning process are dependent upon the inter-object similarity and imply that there is a difference in the neuronal representation for object recognition when using objects with high and low inter-object similarity.     

Prior experience of rotation is not required for recognizing objects seen from different angles

An object viewed from different angles can be recognized and distinguished from similar distractors after the viewer has had experience watching it rotate. It has been assumed that as an observer watches the rotation, separate representations of individual views become associated with one another. However, we show here that once monkeys learned to discriminate individual views of objects, they were able to recognize objects across rotations up to 60 degrees , even though there had been no opportunity to learn the association between different views. Our results suggest that object recognition across small or medium changes in viewing angle depends on features common to similar views of objects.     

Effects of perceptual learning in visual backward masking on the responses of macaque inferior temporal neurons

Learning is critical for fast and efficient object recognition in primates. To understand the neuronal correlates of behavioral improvements due to training, we recorded the responses of single neurons in the inferior temporal (IT) cortex of monkeys that were trained to recognize briefly presented, backward-masked objects. First we investigated training effects that are specific to the objects shown during training and that do not transfer to untrained objects. Only one of two monkeys tested showed object-specific training effects at the behavioral level, and only this monkey showed a transient object-specific increase in object selectivity for trained compared with untrained backward-masked objects. However, in each monkey a substantial part of the training effect transferred to untrained objects. To investigate the neural correlates of these object-independent training effects, we compared the neural responses to masked objects in trained monkeys to the responses in untrained monkeys. Training was associated with a reduction of the responses to the irrelevant masking patterns. These findings suggest that extensive training in recognizing backward-masked objects results in neural changes that reduce IT responses to the interfering irrelevant masking patterns and enhance the processing of the relevant objects.     

"I've seen it all before": explaining age-related impairments in object recognition. Theoretical comment on Burke et al. (2010)

Animal models are useful in elucidating the neural basis of age-related impairments in cognition. Burke, Wallace, Nematollahi, Uprety, and Barnes (2010) tested young and aged rats in several different protocols to measure object recognition memory and found that object recognition deficits in aged rats were consistent with these rats behaving as if novel objects were familiar, rather than familiar objects being treated as novel (that is, forgotten). A similar pattern of behavior has been observed in young rats with perirhinal cortex lesions. Moreover, age-related impairments in object recognition were uncorrelated with deficits in spatial learning in the water maze, a task that requires the integrity of the hippocampus and is also reliably impaired in aged rats. Taken together, these findings support functional specialization of structures within the medial temporal lobe "memory system," as well as the independence of age-related deficits in different cognitive domains. They also potentially form a foundation for neurobiological study of age-related impairments in perirhinal cortex function.     

Multiple levels of visual object constancy revealed by event-related fMRI of repetition priming

We conducted two event-related functional magnetic resonance imaging (fMRI) experiments to investigate the neural substrates of visual object recognition in humans. We used a repetition-priming method with visual stimuli recurring at unpredictable intervals, either with the same appearance or with changes in size, viewpoint or exemplar. Lateral occipital and posterior inferior temporal cortex showed lower activity for repetitions of both real and non-sense objects; fusiform and left inferior frontal regions showed decreases for repetitions of only real objects. Repetition of different exemplars with the same name affected only the left inferior frontal cortex. Crucially, priming-induced decreases in activity of the right fusiform cortex depended on whether the three-dimensional objects were repeated with the same viewpoint, regardless of whether retinal image size changed; left fusiform decreases were independent of both viewpoint and size. These data show that dissociable subsystems in ventral visual cortex maintain distinct view-dependent and view-invariant object representations.     

Color sensitivity of cells responsive to complex stimuli in the temporal cortex

The inferotemporal (IT) cortex of the monkey lies at the head of the ventral visual pathway and is known to mediate object recognition and discrimination. It is often assumed that color plays a minor role in the recognition of objects and faces because discrimination remains highly accurate with black-and-white images. Furthermore it has been suggested that for rapid presentation and reaction tasks, object classification may be based on a first wave of feedforward visual information, which is coarse and achromatic. The fine detail and color information follows later, allowing similar stimuli to be discriminated. To allow these theories to be tested, this study investigates whether the presence of color affects the response of IT neurons to complex stimuli, such as faces, and whether color information is delayed with respect to information about stimulus form in these cells. Color, achromatic, and false-color versions of effective stimuli were presented using a rapid serial visual presentation paradigm, and responses recorded from single cells in IT of the adult monkey. Achromatic images were found to evoke significantly reduced responses compared with color images in the majority of neurons (70%) tested. Differential activity for achromatic and colored stimuli was evident from response onset with no evidence to support the hypothesis that information about object color is delayed with respect to object form. A negative correlation (P < 0.01) was found between cell latency and color sensitivity, with the most color-sensitive cells tending to respond earliest. The results of this study suggest a strong role for color in familiar object recognition and provide no evidence to support the idea of a first wave of form processing in the ventral stream based on purely achromatic information.     

Object-centered encoding by face-selective neurons in the cortex in the superior temporal sulcus of the monkey

Summary Neurophysiological studies have shown that some neurons in the cortex in the superior temporal sulcus and in the inferior temporal cortex respond to faces. To determine if some face responsive neurons encode stimuli in an object-centered coordinate system rather than a viewer-centered coordinate system, a large number of neurons were tested for sensitivity to head movement in 3 macaque monkeys. Ten neurons responded only when a head undergoing rotatory movements was shown. All of these responded to a particular movement independently of the orientation of the moving head in relation to the viewer, maintaining specificity even when the moving head was inverted or shown from the back, thereby reversing viewer-centered movement vectors. This was taken as evidence that the movement was encoded in object-centered coordinates. In tests of whether there are neurons in this area which respond differently to the faces of different individuals relatively independently of viewing angle, it was found that a further 18 neurons responded more to one static face than another across different views. However, for 16 of these 18 cells there was still some modulation of the neuronal response with viewing angle. These 16 neurons thus did not respond perfectly in relation to the object shown independently of viewing angle, and may represent an intermediate stage between a viewercentered and an object-centered representation. In the same area as these neurons, other cells were found which responded on the basis of viewercentered coordinates. These neurophysiological findings provide evidence that some neurons in the inferior temporal visual cortex respond to faces (or heads) on the basis of object-centered coordinates, and that others have responses which are intermediate between object-centered and viewer-centered representations. The results are consistent with the hypothesis that object-centered representations are built in the inferior temporal visual cortex.     

Recollection is delayed under changed viewing conditions: A graded effect on the latency of the late posterior component

Object recognition is a central human ability. In everyday life, the conditions under which objects have to be recognized are usually not perfect. Often, viewing conditions change in between two encounters with an object; typical are changes in illumination or in the object‐observer distance. With such changes, object recognition sometimes feels slightly delayed. We examined this phenomenon empirically by measuring the latency of the well‐established electrophysiological correlate of recollection, the late posterior component (LPC), in an object‐recognition task. Although the cognitive processes underlying successful recognition are well examined, thus far the consequences of changed viewing conditions on the timing of these processes have not been investigated. The ERP technique is well suited for investigating this question, because it allows differentiating between processes contributing to recognition times (in particular, recollection from familiarity as indexed by the FN400 component) and measuring their time course with high temporal precision. In the present study, participants' task was to differentiate previously studied (old) objects from a set of new objects. Viewing conditions for old objects changed slightly, changed strongly, or remained identical between learning and test. We found that the latency of the LPC in response to an old object was delayed whenever viewing conditions changed. Moreover, this delay in LPC latency scaled with the size of the change. These effects were absent for the FN400. This is the first examination of effects of changes in viewing conditions on the latency of recollection and the first dissociation of FN400 and LPC latencies.     

Localization of grasp representations in humans by PET: 1. Observation versus execution

Positron emission tomography (PET) was used to localize brain regions that are active during the observation of grasping movements. Normal, right-handed subjects were tested under three conditions. In the first, they observed grasping movements of common objects performed by the experimenter. In the second, they reached and grasped the same objects. These two conditions were compared with a third condition consisting of object observation. On the basis of monkey data, it was hypothesized that during grasping observation, activations should be present in the region of the superior temporal sulcus (STS) and in inferior area 6. The findings in humans demonstrated that grasp observation significantly activates the cortex of the middle temporal gyrus including that of the adjacent superior temporal sulcus (Brodmann's area 21) and the caudal part of the left inferior frontal gyrus (Brodmann's area 45). The possible functional homologies between these areas and the monkey STS region and frontal area F5 are discussed.     

The effect of catecholaminergic depletion within the prelimbic and infralimbic medial prefrontal cortex on recognition memory for recency, location, and objects

There is good evidence that the medial prefrontal cortex (mPFC) is involved in different aspects of recognition memory. However, the mPFC is a heterogeneous structure, and the contribution of the prelimbic (PL) and infralimbic (IL) cortices to recognition memory has not been investigated. Similarly, the role of different neuromodulators within the mPFC in these processes is poorly understood. To this end, we tested animals with 6-hydroxydopamine (6-OHDA) lesions of the PL and IL mPFC on three tests of object recognition memory that required judgments about recency, object location, and object identity. In the recency task, lesions to both PL and IL severely impaired animals' ability to differentiate between old (earlier presented) and recently presented familiar objects. Relative to sham and PL animals, the IL lesion also disrupted performance on the object location task. However, both lesions left novel object recognition intact. These data confirm previous reports that the mPFC is not required for discriminations based on the relative familiarity of individual objects. However, these results demonstrate that catecholamines within the PL cortex are crucial for relative recency judgments and suggest a possible role for neural processing within the IL in the integration of information about object location.     

The influence of training views on infants' long-term memory for simple 3D shapes

This investigation explored infants' ability to retrieve a memory for a simple 3D shape from a novel view following a 24-hr delay. Tests of memory for shape in infancy have typically used extremely short delay intervals between familiarization and test in examining the ability to equate between substantially different views of a 3D object. The current study used longer delays to assess the content of a long-term memory representation. Infants 3-4 months of age learned to kick to move a mobile displaying a simple 3D shape (brick or cylinder). Results of three experiments show that infants can recognize 3D shapes in a novel viewpoint across a 24-hr delay, provided that experience with a sufficiently wide range of views is available during training. The results suggest a capacity for the perception of 3D shape that enables access, across long delays, to a memory representation of sufficient detail that discrimination between two simple shapes (i.e., a cylinder and a brick) is possible. The results suggest that this representation is of a sufficiently abstract nature that perception of the 3D form of the object, independent of the changes in specific features accompanying changes in viewpoint, is also possible. This finding suggests that infants, like adults, possess a functional memory system for the distal shape of simple 3D objects, and can transfer training to a novel view using long-term memory, but that this ability is not as strong as in the mature system. These results have implications for the development of shape perception and for theories of object recognition in general.     

Is face recognition not so unique after all?

In monkeys, a number of different neocortical as well as limbic structures have cell populations that respond preferentially to face stimuli. Face selectivity is also differentiated within itself: Cells in the inferior temporal and prefrontal cortex tend to respond to facial identity, others in the upper bank of the superior temporal sulcus to gaze directions, and yet another population in the amygdala to facial expression. The great majority of these cells are sensitive to the entire configuration of a face. Changing the spatial arrangement of the facial features greatly diminishes the neurons' response. It would appear, then, that an entire neural network for faces exists which contains units highly selective to complex configurations and that respond to different aspects of the object "face." Given the vital importance of face recognition in primates, this may not come as a surprise. But are faces the only objects represented in this way? Behavioural work in humans suggests that nonface objects may be processed like faces if subjects are required to discriminate between visually similar exemplars and acquire sufficient expertise in doing so. Recent neuroimaging studies in humans indicate that level of categorisation and expertise interact to produce the specialisation for faces in the middle fusiform gyrus. Here we discuss some new evidence in the monkey suggesting that any arbitrary homogeneous class of artificial objects-which the animal has to individually learn, remember, and recognise again and again from among a large number of distractors sharing a number of common features with the target-can induce configurational selectivity in the response of neurons in the visual system. For all of the animals tested, the neurons from which we recorded were located in the anterior inferotemporal cortex. However, as we have only recorded from the posterior and anterior ventrolateral temporal lobe, other cells with a similar selectivity for the same objects may also exist in areas of the medial temporal lobe or in the limbic structures of the same "expert" monkeys. It seems that the encoding scheme used for faces may also be employed for other classes with similar properties. Thus, regarding their neural encoding, faces are not "special" but rather the "default special" class in the primate recognition system.     

Age-associated deficits in pattern separation functions of the perirhinal cortex: a cross-species consensus

Normal aging causes a decline in object recognition. Importantly, lesions of the perirhinal cortex produce similar deficits and also lead to object discrimination impairments when the test objects share common features, suggesting that the perirhinal cortex participates in perceptual discrimination. The current experiments investigated the ability of young and aged animals to distinguish between objects that shared features with tasks with limited mnemonic demands. In the first experiment, young and old rats performed a variant of the spontaneous object recognition task in which there was a minimal delay between the sample and the test phase. When the test objects did not share any features ("Easy" perceptual discrimination) both young and aged rats correctly identified the novel object. When the test objects contained overlapping features, however, only the young rats showed an exploratory preference for the novel object. In Experiment 2, young and aged monkeys were tested on an object discrimination task. When the object pairs were dissimilar, both the young and aged monkeys learned to select the rewarded object quickly. In contrast, when LEGOs? were used to create object pairs with overlapping features, the aged monkeys took significantly longer than did the young animals to learn to discriminate between the rewarded and the unrewarded object. Together, these data indicate that behaviors requiring the perirhinal cortex are disrupted in advanced age, and suggest that at least some of these impairments may be explained by changes in high-level perceptual processing in advanced age.     

IS FACE RECOGNITION NOT SO UNIQUE AFTER ALL?

In monkeys, a number of different neocortical as well as limbic structures have cell populations that respond preferentially to face stimuli. Face selectivity is also differentiated within itself: Cells in the inferior temporal and prefrontal cortex tend to respond to facial identity, others in the upper bank of the superior temporal sulcus to gaze directions, and yet another population in the amygdala to facial expression. The great majority of these cells are sensitive to the entire configuration of a face. Changing the spatial arrangement of the facial features greatly diminishes the neurons' response. It would appear, then, that an entire neural network for faces exists which contains units highly selective to complex configurations and that respond to different aspects of the object “face.” Given the vital importance of face recognition in primates, this may not come as a surprise. But are faces the only objects represented in this way? Behavioural work in humans suggests that nonface objects may be processed like faces if subjects are required to discriminate between visually similar exemplars and acquire sufficient expertise in doing so. Recent neuroimaging studies in humans indicate that level of categorisation and expertise interact to produce the specialisation for faces in the middle fusiform gyrus. Here we discuss some new evidence in the monkey suggesting that any arbitrary homogeneous class of artificial objects—which the animal has to individually learn, remember, and recognise again and again from among a large number of distractors sharing a number of common features with the target—can induce configurational selectivity in the response of neurons in the visual system. For all of the animals tested, the neurons from which we recorded were located in the anterior inferotemporal cortex. However, as we have only recorded from the posterior and anterior ventrolateral temporal lobe, other cells with a similar selectivity for the same objects may also exist in areas of the medial temporal lobe or in the limbic structures of the same “expert” monkeys. It seems that the encoding scheme used for faces may also be employed for other classes with similar properties. Thus, regarding their neural encoding, faces are not “special” but rather the “default special” class in the primate recognition system.     

Graded size sensitivity of object-exemplar-evoked activity patterns within human LOC subregions

A central issue for understanding visual object recognition is how the cortical hierarchy represents incoming sensory information and transforms it across successive processing stages. The format of object representation in the human brain has thus far mostly been studied using adaptation paradigms because the neuronal layout of object selectivities was thought to be beyond the resolution of conventional functional MRI (fMRI). Recently, however, multivariate pattern recognition succeeded in discriminating fMRI responses of object-selective cortex to different object exemplars within a given category. Here, we use increased spatial fMRI resolution to explore size sensitivity and tolerance to size change of response patterns evoked by object exemplars across a range of three sizes. Results from Support Vector Classification on responses of the human lateral occipital complex (LOC) show that discrimination of size (for a given object) and discrimination of objects across changes in size depended on the amount of size difference. Even across the largest amount of size change, accuracy for generalization was still significant in LOC, whereas the same comparison was at chance performance in early visual (calcarine) cortex. Analyzing subregions, we further found an anterior-posterior gradient in the degree of size sensitivity and size generalization within the posterior-dorsal and anterior-ventral parts of LOC. These results speak against fully size-invariant representation of object information in human LOC and are hence congruent with findings in monkeys showing object identity and size information in population activity of inferotemporal cortex. Moreover, these results provide evidence for a fine-grained functional heterogeneity within human LOC beyond the commonly used LO/fusiform subdivision.     

Dominance of the left oblique view in activating the cortical network for face recognition

Faces in portraits are often depicted from the left 3/4 view (an oblique view of the face that is intermediate between the frontal view and left profile). Here, we used functional magnetic resonance imaging (fMRI) to show that, compared with photographs of right 3/4 views of familiar faces, photographs of left 3/4 views of the same faces elicited stronger neural responses in the right middle occipital/inferior parietal cortex, and right inferior frontal gyrus; which are known to be involved in face recognition. By contrast, there was no differential activation in the temporal cortex including the superior temporal sulcus and fusiform gyrus, which are thought to process face-related visual stimuli at a stage that precedes recognition. We suggest that the preference for the left 3/4 view of faces was produced at a later stage of facial information processing that involves attention or memory retrieval.     

Size constancy in rhesus monkeys: Effects of pulvinar,prestriate, and inferotemporal lesions

Summary The present study tested the theory that inferotemporal cortex integrates 1) distance information transmitted via superior colliculus-pulvinar afferents, with 2) form information transmitted via striate-prestriate cortex afferents (Gross, 1973a, 1973b). Monkeys were trained to choose the larger of two objects, independent of distance, to obtain a reward. Based on the integration theory, the following predictions concerning this size constancy discrimination were made: 1) monkeys with pulvinar lesions, unable to code distance, should be impaired and adopt strategies based on retinal image size; and 2) monkeys with prestriate lesions, unable to code retinal image size, should be impaired and adopt strategies based on distance. Contrary to these predictions, pulvinar lesions produced no deficit; and although prestriate lesions did produce an impairment, it was due to a failure to code distance in assessing the true size of the object. Thus, monkeys with prestriate lesions consistently responded to retinal image size instead of object size. Replicating an earlier report (Humphrey and Weiskrantz, 1969), inferotemporal lesions also produced an impairment; however, errors made by monkeys with inferotemporal lesions were random and could not be attributed to any consistent strategy. All monkeys reacquired the discrimination postoperatively, indicating that there are multiple mechanisms available to the brain-damaged animal for the perception of size constancy.