Peculiarities of the Activity of the Brain Structures of People with Schizophrenia During the Categorization of Objects of Animate and Inanimate Nature

In this Chapter the theoretical framework of the knowledge implementation method is presented. In Chapter 8 the knowledge implementation approach is applied for learning of the knowledge and skills of the different categories of objects. The shape understanding system (SUS) operates in two main modes, the learning mode and understanding mode. Learning and understanding are complementary processes. The ability of SUS to understand depends on the effectiveness of learning process and, in turn, learning of the new knowledge depends on the SUS ability to understand. The knowledge implementation is based on assumption that a system to be able to understand needs to learn and learned knowledge needs to be fully understood, that means, that there is a relation among specific learned facts stored in memory. The knowledge implementation is concerned with learning of the visual and the non-visual knowledge from the different categories of objects. The category of the visual objects was described in Chapter 6. The category of the sensory objects and the category of the text objects are related to the category of the visual objects and these categories will be defined in this Chapter. These categories are described in the context of learning and understanding of the visual object, the sensory object and the text object. Proposed new learning methods, that are part of the knowledge implementation approach, are designed to learn the knowledge of an object from the selected category such as the category of the visual objects, the category of the sensory objects or the category of the text objects. In Section 7.1 understanding and learning of the knowledge of the visual object is presented, in Section 7.2 understanding and learning of the knowledge of the sensory object is presented and in Section 7.3 understanding and learning of the knowledge of the text object is presented. Understanding and learning of the knowledge of the visual object, described in Section 7.1, is considered as acquiring the complex perceptual skills by SUS. Acquiring the complex perceptual skills by SUS, needed in perceiving of the world, is connected with implementation of the sophisticated image processing algorithms and learning of the complex patterns of the visual reasoning sequences. The visual knowledge is learned based on the sample of the visual objects that represent the category of visual objects selected for learning and utilization of the different forms of the visual abstraction. The important forms of the visual abstraction are generalization and specialization described in Sections 7.1.1 and 7.1.2. Learning of the visual knowledge is called the categorical learning and the short description of this method was presented in [1]. In this Section the new developed categorical learning techniques: learning by the small alternation of the visual object (LSA), learning from the simple to complex (LSC), learning by simplification of the complex object (LSCO), learning from parts (LP), and learning parts decomposition (LPD), are described. The new developed categorical learning techniques are applied to learn the knowledge of the selected ontological categories. Learning of the knowledge of the selected ontological categories will be described in Chapter 8.

Evidence suggests that cortical minicolumns are reduced in size and increased in number in individuals with autism spectrum disorder (ASD), especially in the dorsolateral prefrontal cortex (DLPFC). More specifically minicolumns in individuals with ASD are narrower and contain less peripheral, neuropil space; this may cause an increase in the ratio of cortical excitation to inhibition and adversely affect the functional distinctiveness of minicolumnar activation. A lack of cortical inhibition may cause signal/sensory amplification which can impair functioning, raise physiological stress, and adversely affect social interaction in patients with ASD. Additionally, the DLPFC forms a circuit interconnected with many areas of cortex (e.g., anterior cingulate, orbitofrontal) and is involved in selecting a possible range of responses while suppressing inappropriate ones. Low-frequency (:'SlHz) repetitive transcranial magnetic stimulation (rTMS) has been shown to increase inhibition of stimulated cortex by the activation of inhibitory circuits. The baseline hypothesis was that individuals with ASD would show electroencephalopgrahic (EEG) and event-related potential (ERP) evidence of amplified cortical activity at early and late stages of visual processing as well as impaired indices of selective attention. The second hypothesis was that low-frequency rTMS would reduce augmented cortical responses at early stage and late stages of visual processing and improve selective attention and behavior in ASD. The baseline findings indicate both ERP and evoked gamma activity are amplified and indiscriminative in ASD at early stages of visual processing which may reflect decreased 'signal to noise' due to decreased cortical inhibitory processing. Additionally, individuals with ASD showed evidence of compromised selective attention, and had a significantly higher rate of motor response errors. After low-frequency rTMS individuals with ASD showed significant reductions in augmented ERP responses at very early stages of visual processing and showed significant improvement in discriminatory EEG gamma activity. There was also evidence of improved ERP indices of selective attention and significant reductions in irritability and repetitive behavior. TMS has the potential to become an important therapeutic tool in ASD treatment and has shown significant benefits in treating core symptoms of ASD with few, if any side effects.

Archerfish hunt by shooting a jet of water at aerial targets, a behavior used to study their visual processing by presenting a set of images on a screen above the water tank and observing the behavioral response. Building on this unique behavior, it was recently shown that archerfish can be trained to distinguish between different object categories by generalizing from examples. Analysis of the archerfish’s behavior revealed that the fish visual system relies on a small set of visual features for categorization and is more sensitive to object contours than to textures. To understand the neural basis of this object recognition, we investigated the neural representation of features and objects in the archerfish optic tectum using recording of single cells. We found that, although the optic tectum is an early stage of visual processing, a small population of neurons in this region contains information about the object category. This contrasts with the primate visual system, where the representation of objects emerges only at later stages of visual processing. These results suggest that early-stage feature extraction and object categorization in archerfish might represent a form of specialized visual processing. This contributes to a broader understanding of visual processing across taxa.

This study aimed to explore the differential role of the frontoparietal network in processing different visual object categories, matched for difficulty level, during a 1-back paradigm. To achieve this goal, we first mapped the effort-related frontoparietal saliency network, by contrasting activation elicited by face, object, place, body and verbal stimulus categories, which were matched for performance level, and speed of processing, with difficult scrambled stimuli. We then computed the weight of object predictors on that specific network, using an independent orthogonal analysis. Overall, our results demonstrated that face (and to some extent also places) stimuli were associated with lower processing load in regions of the frontoparietal network comparing to other visual categories, suggesting that face/place processing does require to a much smaller extent the recruitment of the frontoparietal control network than any other object categories. Thus, face detection and place detection seem to be routed in specific neuronal systems that readily encode the holistic nature of this type of objects. We conclude that the more limited recruitment of frontoparietal networks reflects the automaticity of face and place processing and their smaller dependence on general capacity limits.

In this paper, we deal with two research issues: the automation of visual attribute identification and semantic relation learning between visual attributes and object categories. The contribution is two-fold, firstly, we provide uniform framework to reliably extract both categorical attributes and depictive attributes. Secondly, we incorporate the obtained semantic associations between visual attributes and object categories into a text-based topic model and extract descriptive latent topics from external textual knowledge sources. Specifically, we show that in mining natural language descriptions from external knowledge sources, the relation between semantic visual attributes and object categories can be encoded as Must-Links and Cannot-Links, which can be represented by Dirichlet-Forest prior. To alleviate the workload of manual supervision and labeling in image categorization process, we introduce a semi-supervised training framework using soft-margin semi-supervised SVM classifier. We also show that the large-scale image categorization results can be significantly improved by combining automatically acquired visual attributes. Experimental results show that the proposed model achieves better ability in describing object-related attributes and makes the inferred latent topics more descriptive.

The general objective of this work is to develop a description language for constructive 3D boundary representation [5] of neuroanatomical structures and connectivity at various levels of granularity (from coarse-resolution solids to fine meshes). This approach is motivated by a desire to capture regularities in neural circuitry as revealed by neuro-architectonic studies [4], while at the same time to explore hypotheses about anatomic variability of various origins, such as experimental uncertainties, cross species scaling factors, individual differences, and assess them for their impact on connectivity, and ultimately on network dynamics. Boundary representation is a general approach to describe 3D objects solely by their surface. Boundary representations consist of topological objects and their concrete geometrical representations in terms of enclosing boundaries. Topological elements include vertices, edges and faces, with corresponding geometrical elements being points, curves, and surfaces. The relationships between topological elements in a structure are expressed by means of a graph of topological connections. The description language is being implemented on top of the GNU Triangulated Surface library [3], and provides the ability to: • Specify compound topological objects with parametric geometry; • Specify geometric parameters for the instantiation of topological objects, such as coordinates for placement, or probability distributions for random placement of a group of identical objects; • Specify individual coordinate systems for different cell populations; • Define categories of topological objects, such as stellate, basket and Golgi cells, which may be part of a morphological continuum; • Define rules for connectivity between different categories of objects. We present the anisotropic cerebellar circuitry as a case study, and define boundary representations of the arrangement of Purkinje and Golgi cells in the cerebellar cortex. We use the hexagonal grid pattern suggested by Palkovits et al. in [1,2], while allowing for small variability in the placement of cells within a hexagon. The implementation of the language instantiates the topological objects, computes the intersections of the resulting surfaces, and given connectivity rules for the different categories of objects, computes the potential synaptic connectivity (producing graph theoretical measures, as well as connectivity histograms), and ultimately aims to generate connectivity descriptions for the NEURON simulator.

Mental rotation and object categorization share a common network of prefrontal and dorsal and ventral regions of posterior cortex

fMR-adaptation reveals a distributed representation of inanimate objects and places in human visual cortex

Objects belonging to different categories evoke reliably different fMRI activity patterns in human occipitotemporal cortex, with the most prominent distinction being that between animate and inanimate objects. An unresolved question is whether these categorical distinctions reflect category-associated visual properties of objects or whether they genuinely reflect object category. Here, we addressed this question by measuring fMRI responses to animate and inanimate objects that were closely matched for shape and low-level visual features. Univariate contrasts revealed animate- and inanimate-preferring regions in ventral and lateral temporal cortex even for individually matched object pairs (e.g., snake-rope). Using representational similarity analysis, we mapped out brain regions in which the pairwise dissimilarity of multivoxel activity patterns (neural dissimilarity) was predicted by the objects' pairwise visual dissimilarity and/or their categorical dissimilarity. Visual dissimilarity was measured as the time it took participants to find a unique target among identical distractors in three visual search experiments, where we separately quantified overall dissimilarity, outline dissimilarity, and texture dissimilarity. All three visual dissimilarity structures predicted neural dissimilarity in regions of visual cortex. Interestingly, these analyses revealed several clusters in which categorical dissimilarity predicted neural dissimilarity after regressing out visual dissimilarity. Together, these results suggest that the animate-inanimate organization of human visual cortex is not fully explained by differences in the characteristic shape or texture properties of animals and inanimate objects. Instead, representations of visual object properties and object category may coexist in more anterior parts of the visual system.

Early and late stages of visual processing in individuals in prodromal state and first episode schizophrenia: An ERP study

The recently introduced angular integral of the Radon transform (aniRT) seems to be a good candidate as a feature vector used in categorization of visual objects in a rotation invariant fashion. We investigate application of aniRT in situations when the number of objects is significant, for example, Chinese characters. Typically, the aniRT feature vector spans the diagonal of the visual object. We show that a subset of the full aniRT vector delivers a good categorization results in a timely manner.

The brain mechanisms that integrate the separate features of sensory input into a meaningful percept depend upon the prior experience of interaction with the object and differ between categories of objects. Recent studies using representational similarity analysis (RSA) have characterized either the spatial patterns of brain activity for different categories of objects or described how category structure in neuronal representations emerges in time, but never simultaneously. Here we applied a novel, region-based, multivariate pattern classification approach in combination with RSA to magnetoencephalography data to extract activity associated with qualitatively distinct processing stages of visual perception. We asked participants to name what they see whilst viewing bitonal visual stimuli of two categories predominantly shaped by either value-dependent or sensorimotor experience, namely faces and tools, and meaningless images. We aimed to disambiguate the spatiotemporal patterns of brain activity between the meaningful categories and determine which differences in their processing were attributable to either perceptual categorization per se, or later-stage mentalizing-related processes. We have extracted three stages of cortical activity corresponding to low-level processing, category-specific feature binding, and supra-categorical processing. All face-specific spatiotemporal patterns were associated with bilateral activation of ventral occipito-temporal areas during the feature binding stage at 140–170 ms. The tool-specific activity was found both within the categorization stage and in a later period not thought to be associated with binding processes. The tool-specific binding-related activity was detected within a 210–220 ms window and was located to the intraparietal sulcus of the left hemisphere. Brain activity common for both meaningful categories started at 250 ms and included widely distributed assemblies within parietal, temporal, and prefrontal regions. Furthermore, we hypothesized and tested whether activity within face and tool-specific binding-related patterns would demonstrate oppositely acting effects following procedural perceptual learning. We found that activity in the ventral, face-specific network increased following the stimuli repetition. In contrast, tool processing in the dorsal network adapted by reducing its activity over the repetition period. Altogether, we have demonstrated that activity associated with visual processing of faces and tools during the categorization stage differ in processing timing, brain areas involved, and in their dynamics underlying stimuli learning.

Event Abstract Back to Event Comparison of early visual ERP components in different object categories with low inter stimulus perceptual variance Sinem Kara1*, M. Cursi1, N. Amato1, A. Inuggi1, S. Velikova1, G. Comi1 and L. Leocani1 1 University Hospital San Raffaele, Department of Neurology and Clinical Neurophysiology, INSPE, Italy Background: In a recent study of Thierry et al. (2007) it was claimed that category effects previously reported in the N170 range may be due in part to uncontrolled inter stimulus perceptual variance (ISPV) differences between experimental conditions. Since neuroimaging studies of visual face and object recognition generally use full front views of faces contrasted with pictures of other objects presented in a variety of sizes and spatial layouts, category contrasts typically involve implicit comparisons of low and high ISPV conditions. They hypothesized that the face selectivity of the N170 might be an artifact driven by ISPV differences which are basically eccentricity, orientation and size. Aim: We examined the P100 and N170 waves with respect to low ISPV condition in different object categories considering that if N170 is sensitive to ISPV but not to the object categories then its amplitude should be relatively insensitive to different object categories with low ISPV. In addition, we also examined the progressive averaging of pictures of our object categories with the aim of evaluating template effect on P100 and N170 waves. Our full front face pictures’ average was still recognizable as a face which might have created a stronger template effect on the working memory even if the faces were presented in a pseudo-random order. Therefore for comparison we preferred cars and fruits so that the average of cars was still recognizable as a car like it was in faces, while the average of fruits wasn’t recognizable as a category but formed a big round border (because all the fruits were chosen from big round fruits like melon and apple). Methods: 15 healthy subjects (8 females, mean age: 27+/-3) were confronted with a visual oddball design, in which they had to detect, as quickly as possible, deviants, a pineapple, amongst a train of standard stimuli with low ISPV (faces, cars and fruits). Recordings were done with 30 channel EEG BrainVision recorder. Results: No significant difference was found amongst the object categories for the P100 latencies and amplitudes and the N170 amplitudes. The only significant difference was seen for the N170 latencies in which the N170 latency of faces was significantly shorter than that of cars and fruits. Conclusion: We saw no difference among the groups for P100 and N170 amplitudes in terms of template effect on working memory. Also we found no measurable effect of object category (face, car and fruits) on both P100 and N170 peak amplitudes. Although our findings for N170 amplitude confirmed the study of Thierry et al about the insensitivity of N170 amplitudes for faces, our data for P100 wave were in conflict with their results about the P100 amplitude being sensitive for object categorization.

In open-ended domains, robots must continuously learn new object categories. When the training sets are created offline, it is not possible to ensure their representativeness with respect to the object categories and features the system will find when operating online. In the Bag of Words model, visual codebooks are usually constructed from training sets created offline. This might lead to non-discriminative visual words and, as a consequence, to poor recognition performance. This paper proposes a visual object recognition system which concurrently learns in an incremental and online fashion both the visual object category representations as well as the codebook words used to encode them. The codebook is defined using Gaussian Mixture Models which are updated using new object views. The approach contains similarities with the human visual object recognition system: evidence suggests that the development of recognition capabilities occurs on multiple levels and is sustained over large periods of time. Results show that the proposed system with concurrent learning of object categories and codebooks is capable of learning more categories, requiring less examples, and with similar accuracies, when compared to the classical Bag of Words approach using codebooks constructed offline.

Recording synchronous data from EEG and eye-tracking provides a unique methodological approach for measuring the sensory and cognitive processes of overt visual search. Using this approach we obtained fixation related potentials (FRPs) during a guided visual search task specifically focusing on the lambda and P3 components. An outstanding question is whether the lambda and P3 FRP components are influenced by concurrent task demands. We addressed this question by obtaining simultaneous eye-movement and electroencephalographic (EEG) measures during a guided visual search task while parametrically modulating working memory load using an auditory N-back task. Participants performed the guided search task alone, while ignoring binaurally presented digits, or while using the auditory information in a 0, 1, or 2-back task. The results showed increased reaction time and decreased accuracy in both the visual search and N-back tasks as a function of auditory load. Moreover, high auditory task demands increased the P3 but not the lambda latency while the amplitude of both lambda and P3 was reduced during high auditory task demands. The results show that both early and late stages of visual processing indexed by FRPs are significantly affected by concurrent task demands imposed by auditory working memory.