High level object recognition without an anterior inferior temporal lobe

Reversible Inactivation of Different Millimeter-Scale Regions of Primate IT Results in Different Patterns of Core Object Recognition Deficits

Extensive research suggests that the inferior temporal (IT) population supports visual object recognition behavior. However, causal evidence for this hypothesis has been equivocal, particularly beyond the specific case of face-selective subregions of IT. Here, we directly tested this hypothesis by pharmacologically inactivating individual, millimeter-scale subregions of IT while monkeys performed several core object recognition subtasks, interleaved trial-by trial. First, we observed that IT inactivation resulted in reliable contralateral-biased subtask-selective behavioral deficits. Moreover, inactivating different IT subregions resulted in different patterns of subtask deficits, predicted by each subregion's neuronal object discriminability. Finally, the similarity between different inactivation effects was tightly related to the anatomical distance between corresponding inactivation sites. Taken together, these results provide direct evidence that the IT cortex causally supports general core object recognition and that the underlying IT coding dimensions are topographically organized.

Behavioral Differences in the Upper and Lower Visual Hemifields in Shape and Motion Perception

Perceptual accuracy is known to be influenced by stimuli location within the visual field. In particular, it seems to be enhanced in the lower visual hemifield (VH) for motion and space processing, and in the upper VH for object and face processing. The origins of such asymmetries are attributed to attentional biases across the visual field, and in the functional organization of the visual system. In this article, we tested content-dependent perceptual asymmetries in different regions of the visual field. Twenty-five healthy volunteers participated in this study. They performed three visual tests involving perception of shapes, orientation and motion, in the four quadrants of the visual field. The results of the visual tests showed that perceptual accuracy was better in the lower than in the upper visual field for motion perception, and better in the upper than in the lower visual field for shape perception. Orientation perception did not show any vertical bias. No difference was found when comparing right and left VHs. The functional organization of the visual system seems to indicate that the dorsal and the ventral visual streams, responsible for motion and shape perception, respectively, show a bias for the lower and upper VHs, respectively. Such a bias depends on the content of the visual information.

Simple Learned Weighted Sums of Inferior Temporal Neuronal Firing Rates Accurately Predict Human Core Object Recognition Performance

To go beyond qualitative models of the biological substrate of object recognition, we ask: can a single ventral stream neuronal linking hypothesis quantitatively account for core object recognition performance over a broad range of tasks? We measured human performance in 64 object recognition tests using thousands of challenging images that explore shape similarity and identity preserving object variation. We then used multielectrode arrays to measure neuronal population responses to those same images in visual areas V4 and inferior temporal (IT) cortex of monkeys and simulated V1 population responses. We tested leading candidate linking hypotheses and control hypotheses, each postulating how ventral stream neuronal responses underlie object recognition behavior. Specifically, for each hypothesis, we computed the predicted performance on the 64 tests and compared it with the measured pattern of human performance. All tested hypotheses based on low- and mid-level visually evoked activity (pixels, V1, and V4) were very poor predictors of the human behavioral pattern. However, simple learned weighted sums of distributed average IT firing rates exactly predicted the behavioral pattern. More elaborate linking hypotheses relying on IT trial-by-trial correlational structure, finer IT temporal codes, or ones that strictly respect the known spatial substructures of IT ("face patches") did not improve predictive power. Although these results do not reject those more elaborate hypotheses, they suggest a simple, sufficient quantitative model: each object recognition task is learned from the spatially distributed mean firing rates (100 ms) of ∼60,000 IT neurons and is executed as a simple weighted sum of those firing rates. Significance statement: We sought to go beyond qualitative models of visual object recognition and determine whether a single neuronal linking hypothesis can quantitatively account for core object recognition behavior. To achieve this, we designed a database of images for evaluating object recognition performance. We used multielectrode arrays to characterize hundreds of neurons in the visual ventral stream of nonhuman primates and measured the object recognition performance of >100 human observers. Remarkably, we found that simple learned weighted sums of firing rates of neurons in monkey inferior temporal (IT) cortex accurately predicted human performance. Although previous work led us to expect that IT would outperform V4, we were surprised by the quantitative precision with which simple IT-based linking hypotheses accounted for human behavior.

Using automated morphometry to detect associations between ERP latency and structural brain MRI in normal adults

Despite the clinical significance of event-related potential (ERP) latency abnormalities, little attention has focused on the anatomic substrate of latency variability. Volume conduction models do not identify the anatomy responsible for delayed neural transmission between neural sources. To explore the anatomic substrate of ERP latency variability in normal adults using automated measures derived from magnetic resonance imaging (MRI), ERPs were recorded in the visual three-stimulus oddball task in 59 healthy participants. Latencies of the P3a and P3b components were measured at the vertex. Measures of local anatomic size in the brain were estimated from structural MRI, using tissue segmentation and deformation morphometry. A general linear model was fitted relating latency to measures of local anatomic size, covarying for intracranial vault volume. Longer P3b latencies were related to contractions in thalamus extending superiorly into the corpus callosum, white matter (WM) anterior to the central sulcus on the left and right, left temporal WM, the right anterior limb of the internal capsule extending into the lenticular nucleus, and larger cerebrospinal fluid volumes. There was no evidence for a relationship between gray matter (GM) volumes and P3b latency. Longer P3a latencies were related to contractions in left temporal WM, and left parietal GM and WM near the interhemispheric fissure. P3b latency variability is related chiefly to WM, thalamus, and lenticular nucleus, whereas P3a latency variability is not related as strongly to anatomy. These results imply that the WM connectivity between generators influences P3b latency more than the generators themselves do.

Relative sparing of primary auditory cortex in Williams Syndrome

Williams Syndrome (WS) is a neurodevelopment disorder associated with a hemizygous deletion on chromosome 7. WS is characterized with mental retardation, severe visual-spatial deficits, relative language preservation, and excellent facial recognition. Distinctive auditory features include musical ability, heightened sound sensitivity, and specific patterns of auditory evoked potentials. These features have led to the hypothesis that the dorsal forebrain is more affected than the ventral. Previously, we reported primary visual area 17 abnormalities in rostral striate cortex, a region contributing to the dorsal visual pathway. Based on the dorsal-ventral hypothesis, and language and auditory findings, we predicted a more normal histometric picture in auditory area 41. We used an optical dissector method to measure neurons in layers II-VI of area 41 in right and left hemispheres of the same 3 WS and 3 control brains used in the area 17 study. There was a hemisphere by diagnosis interaction in cell packing density (CPD) in layer IV and in cell size in layer III between WS and control brains. Post hoc analysis disclosed in control brains, but not WS, a layer IV left > right asymmetry in CPD, and a layer III left < right asymmetry in cell size. WS brains showed more large neurons bilaterally in layer II and in left layer VI. Histometric alterations in area 41 were less widespread than rostral visual cortex. Also, there was less asymmetry in the WS brain. We interpret layers II and VI differences as reflecting increased limbic connectivity in primary auditory cortex of WS.

Exploring the functional architecture of person recognition system with event-related potentials in a within- and cross-domain self-priming of faces

In this paper, we explored the functional properties of person recognition system by investigating the onset, magnitude, and scalp distribution of within- and cross-domain self-priming effects on event-related potentials (ERPs). Recognition of degraded pictures of famous people was enhanced by a prior exposure to the same person's face (within-domain self-priming) or name (cross-domain self-priming) as compared to those preceded by neutral or unrelated primes. The ERP results showed first that the amplitude of the N170 component to famous face targets was modulated by within- and cross-domain self-priming, suggesting not only that the N170 component can be affected by top-down influences but also that this top-down effect crosses domains. Second, similar to our behavioral data, later ERPs to famous faces showed larger ERP self-priming effects in the within-domain than in the cross-domain condition. In addition, the present data dissociated between two topographically and temporally overlapping priming-sensitive ERP components: the first one, with a strongly posterior distribution arising at an early onset, was modulated more by within-domain priming irrespective whether the repeated face was familiar or not. The second component, with a relatively uniform scalp distribution, was modulated by within- and cross-domain priming of familiar faces. Moreover, there was no evidence for ERP-induced modulations for unfamiliar face targets in the cross-domain condition. Together, our findings suggest that multiple neurocognitive events that are possibly mediated by distinct brain loci contribute to face priming effects.

Spatiotemporal differences between cognitive processes of spatially possible and impossible objects: a high-density electrical mapping study

Differences in cognitive processing between spatially possible and impossible figures were investigated using event-related potentials (ERPs). Two types of figures with identical luminance and equivalent spatial frequency were used as visual stimuli: possible three-dimensional figures (drawn with perspective and existing in the three-dimensional world) and impossible figures (drawn with perspective but not existing in the three-dimensional world). High-density electroencephalographic recording (72 channels) was performed for analysis of ERPs accompanying perception of each figure type; amplitude differences between the conditions were considered neurophysiologic correlates to perceptual differences between possible and impossible objects. Low-resolution brain electromagnetic tomography (LORETA) was used to identify the current source related to the differences. Compared with impossible three-dimensional figures, perception of possible figures showed a significant negative potential increase in the right inferior occipitotemporal region between 350 and 389 ms of latency. The current source was localized to the right fusiform gyrus. The results suggest that right fusiform gyrus is involved in discrimination between spatially possible and impossible objects.

An electrophysiological study of scene effects on object identification

The meaning of a visual scene influences the identification of visual objects embedded in it. We investigated the nature and time course of scene effects on object identification by recording event-related brain potentials (ERPs) and response times (RTs). In three experiments, participants identified objects within a scene that were either semantically congruous (e.g., a pot in a kitchen) or incongruous (e.g., a desk in a river). As expected, RTs were faster for congruous than incongruous objects. The earliest sign of reliable scene congruity effects in the ERPs (greater positivity for congruous pictures between 300 and 500 ms) was around 300 ms. Both the morphology and time course of the N390 scene congruity effect are reminiscent of the N400 sentence congruity effect typically observed in sentence context paradigms, suggesting a functional similarity of the neural processes involved. Overall, these results support theories postulating that visual scenes do not appreciably affect object identification processes before associated semantic information is activated. We speculate that the N390 scene congruity effect reflects the action of visual scene schemata stored in the anterior temporal lobe.

A selective deficit for living things after temporal lobectomy for relief of epileptic seizures

Unilateral left and right temporal lobectomy patients and normal control subjects were tested on confrontation naming, speeded naming, category generation, and category and associate matching tasks. Both groups of patients were disproportionately impaired for living relative to nonliving things in confrontation naming, speeded naming, and category generation. We argue that damage to the temporal lobe impairs lexical retrieval most strongly for living things and that the anterior temporal cortices are convergence zones particularly necessary for retrieving the names of living things.

The role of the posterior parietal cortex in human object recognition: a functional magnetic resonance imaging study

The mechanisms involved in visual object recognition from non-canonical viewpoints were investigated using functional magnetic resonance imaging (fMRI). We used a passive observation task and found three areas activated more strongly in the non-canonical viewing condition compared with the canonical viewing condition. First, it was found that the fusiform gyrus and posterior part of the inferior temporal cortex were involved in the processing of shape information. Next, it was found that the posterior parietal cortex, mainly the superior parietal lobule and the ventral part of premotor area were involved in visuospatial processing and accessing sensorimotor knowledge. These results may indicate that recognition from non-canonical viewpoints is supported by using functional properties of the object, which require more real-time processing for object manipulation.

The neural substrate of picture naming

A PET study of 10 normal males was carried out using the bolus H215O intravenous injection technique to examine the effects of picture naming and semantic judgment on blood flow. In a series of conditions, subjects (1) passively viewed flashing plus signs, (2) noted the occurrence of abstract patterns, (3) named animal pictures, or (4) carried out a semantic judgment on animal pictures. Anticipatory scans were carried out after the subjects were presented with the instructions but before they began the cognitive task, as they were passively viewing plus signs. Our results serve to clarify a number of current controversies regarding the neural substrate of picture naming. The results indicate that the fusiform gyrus is unlikely to be the region where low-level perceptual processing such as shape analysis is undertaken. In fact, our evidence suggests that activation of the fusiform gyrus is most likely related to visual perceptual semantic processing. In addition, the inferior/middle frontal lobe activity observed while performing the picture naming and semantic judgment tasks does not appear to be due to the effects of anticipation or preparation. Furthermore, there appears to be a set of regions (a semantic network) that becomes activated regardless of whether the subjects perform a picture naming or semantic judgment task. Finally, picture naming of animals did not activate either parietal regions or anterior inferior left temporal regions, regardless of what subtraction baseline was used.

Deficits in complex visual perception following unilateral temporal lobectomy

Although human temporal cortex is known to be important for short- and long-term memory, its role in visual perception is not well understood. In this study, we compared the performance of three patients with unilateral temporal lobectomies to that of normal controls on both "simple" and "complex" visual discriminations that did not involve explicit memory components. Two types of complex tasks were tested that involved discriminations secondary to texture segmentation. These were contrasted with simple discriminations using luminance-defined stimuli. Patients showed impaired thresholds only on tasks involving texture segmentation, performing as well as controls when the targets were defined by luminance rather than texture. The minimum stimulus presentation times for threshold performance were also measured for all tasks and found to be elevated in temporal lobectomy patients relative to controls. Although the magnitude of the deficits observed was substantial, loss was equivalent in ipsi- and contra-lesional regions of the visual field. Additional control experiments showed that the patients' perceptual deficits were not due, even in part, to disturbances of basic visual capacities such as acuity and contrast sensitivity. Our results indicate that temporal lobe damage disrupts complex, but not simple, visual discriminations throughout the visual field.

A neural basis for the retrieval of conceptual knowledge

Both clinical reports and systematic neuropsychological studies have shown that patients with damage to selected brain sites develop defects in the retrieval of conceptual knowledge for various concrete entities, leading to the hypothesis that the retrieval of knowledge for entities from different conceptual categories depends on partially segregated large-scale neural systems. To test this hypothesis, 116 subjects with focal, unilateral lesions to various sectors of the telencephalon, and 55 matched controls, were studied with a procedure which required the visual recognition of entities from three categories--unique persons, non-unique animals and non-unique tools. Defective recognition of persons was associated with maximal lesion overlap in right temporal polar region; defective recognition of animals was associated with maximal lesion overlap in right mesial occipital/ventral temporal region and also in left mesial occipital region; and defective recognition of tools was associated with maximal lesion overlap in the occipital-temporal-parietal junction of the left hemisphere. The findings support the hypothesis that the normal retrieval of knowledge for concrete entities from different conceptual domains depends on partially segregated neural systems. These sites may operate as catalysts for the retrieval of the multidimensional aspects of knowledge which are necessary and sufficient for the mental representation of a concept of a given entity.

Neurocomputational bases of object and face recognition

A number of behavioural phenomena distinguish the recognition of faces and objects, even when members of a set of objects are highly similar. Because faces have the same parts in approximately the same relations, individuation of faces typically requires specification of the metric variation in a holistic and integral representation of the facial surface. The direct mapping of a hypercolumn-like pattern of activation onto a representation layer that preserves relative spatial filter values in a two-dimensional (2D) coordinate space, as proposed by C. von der Malsburg and his associates, may account for many of the phenomena associated with face recognition. An additional refinement, in which each column of filters (termed a 'jet') is centred on a particular facial feature (or fiducial point), allows selectivity of the input into the holistic representation to avoid incorporation of occluding or nearby surfaces. The initial hypercolumn representation also characterizes the first stage of object perception, but the image variation for objects at a given location in a 2D coordinate space may be too great to yield sufficient predictability directly from the output of spatial kernels. Consequently, objects can be represented by a structural description specifying qualitative (typically, non-accidental) characterizations of an object's parts, the attributes of the parts, and the relations among the parts, largely based on orientation and depth discontinuities (as shown by Hummel & Biederman). A series of experiments on the name priming or physical matching of complementary images (in the Fourier domain) of objects and faces documents that whereas face recognition is strongly dependent on the original spatial filter values, evidence from object recognition indicates strong invariance to these values, even when distinguishing among objects that are as similar as faces.

Intact Perceptual Priming in a Patient with Damage to the Anterior Inferior Temporal Lobes

We conducted three experiments to examine whether the anterior portion of the inferior temporal (IT) lobe is involved in the processing of visual objects in humans. In monkeys, damage to this region results in severe deficits in perception and in memory for visual objects. Our study was designed to examine both these processes in a patient (DM) with bilateral damage to the anterior portion of the inferior temporal lobe. Neuropsychological examination revealed a significant semantic impairment and a mild deficit in the discrimination of familiar objects from nonobjects. Despite these difficulties, the results of several studies indicated that DM was able to form and retain descriptions of the structure of objects. Specifically, DM showed normal perceptual priming for familiar and novel objects on implicit memory tests, even when the objects were transformed in size and left-right orientation. These results suggest that the anterior IT is not'involved in (1) the storage of pre-existing structural descriptions of known objects, (2) the ability to create new structural descriptions for novel objects, and (3) the ability to compute descriptions that are invariant with respect to changes in size and reflection. Instead, the anterior IT appears to provide the interface between structural descriptions of objects and their meanings.