Why is Real-World Visual Object Recognition Hard?

Progress in understanding the brain mechanisms underlying vision requires the construction of computational models that not only emulate the brain's anatomy and physiology, but ultimately match its performance on visual tasks. In recent years, “natural” images have become popular in the study of vision and have been used to show apparently impressive progress in building such models. Here, we challenge the use of uncontrolled “natural” images in guiding that progress. In particular, we show that a simple V1-like model—a neuroscientist's “null” model, which should perform poorly at real-world visual object recognition tasks—outperforms state-of-the-art object recognition systems (biologically inspired and otherwise) on a standard, ostensibly natural image recognition test. As a counterpoint, we designed a “simpler” recognition test to better span the real-world variation in object pose, position, and scale, and we show that this test correctly exposes the inadequacy of the V1-like model. Taken together, these results demonstrate that tests based on uncontrolled natural images can be seriously misleading, potentially guiding progress in the wrong direction. Instead, we reexamine what it means for images to be natural and argue for a renewed focus on the core problem of object recognition—real-world image variation.     

How Does the Brain Solve Visual Object Recognition?

Mounting evidence suggests that 'core object recognition,' the ability to rapidly recognize objects despite substantial appearance variation, is solved in the brain via a cascade of reflexive, largely feedforward computations that culminate in a powerful neuronal representation in the inferior temporal cortex. However, the algorithm that produces this solution remains poorly understood. Here we review evidence ranging from?individual neurons and neuronal populations to behavior and computational models. We propose that understanding this algorithm will require using neuronal and psychophysical data to sift through many computational models, each based on building blocks of small, canonical subnetworks with a common functional goal.     

Object detection by honeybees: Why do they land on edges?

Behavioural experiments using a variety of experimental situations (Figs. 1, 3, 6, 7, 9) were conducted to investigate the visual cues which bees use in the task of object-ground discrimination. The bees' flight and landing behaviour was video-filmed throughout the experiments. The evaluation of the video data shows that bees trained to find a randomly textured figure raised above a similarly textured ground land mainly on the boundaries of the figure, facing its inner surface (Fig. 2a, b). Bees can also be trained to find a hole, i.e. a low texture viewed through a window cut in a raised texture, but these bees are not attracted to the edges of the hole (Fig. 5a, b). Bees trained to a single edge between a low and a raised random texture land at the edge mainly facing the raised side (Table 1). Bees approaching the edge from the high side cross the edge in most cases without landing on it (Table 1). Bees trained to an edge between 2 striped patterns, one raised above the other, again land on the edge facing the raised pattern, regardless of whether the stripes on the 2 patterns run parallel or perpendicular to each other or to the edge (Fig. 8). In this case, the bees acquire range information by flying in oblique directions with respect to the orientation of the stripes (Fig. 10). All of the results suggest that the edge elicits landings when the bee perceives a local increase in the speed of image motion, signalling an abrupt decrease in range. This is corroborated by the results of further experiments in which artificial motion was used to simulate range differences between the two sides of an edge (Table 2). We conclude that image speed is a powerful cue in range discrimination as well as object detection. Dedicated to G. Adrian Horridge on the occasion of his retirement     

Why Would Anyone Object? An Exploration of Social Aspects of Phytoremediation Acceptability

Referee: Mr. Peter Siebach, Environmental Engineer, U.S. Department of Energy, Chicago Operations Office, 9800 South Cass Avenue, Argonne, IL 60439 Phytoremediation comprises a suite of promising cleanup technologies that use plants to remove or contain contaminants in soil and water. To be deployed, phytoremediation must be both technically and socially acceptable. This article explores the potential social acceptability of phytoremediation options proposed for use at specific sites and describes the conceptual framework that guides our exploration. The framework, called PACT (Public Acceptability of Controversial Technologies), consists of Dialog, Technology, Constituent, and Context dimensions. It posits that remediation decision making is a social process informed by scientific and technical information, rather than a science- or technology-driven process. Although empirical data are scarce, applying PACT shows that a number of issues have the potential to impose conditions on the social acceptability of phytoremediation, and that some issues could lead to outright rejection. Further, because many of these issues concern values and goals, they cannot be resolved simply by providing better or more detailed technical information about phytoremediation. PACT is instructive in showing how even seemingly benign or desirable technologies such as phytoremediation have the potential to generate public controversy, delineating issues in ways that can help lead to their resolution.     

The Cell and Protoplasm as Container,Object, and Substance, 1835–1861

This article revisits the development of the protoplasm concept as it originally arose from critiques of the cell theory, and examines how the term “protoplasm” transformed from a botanical term of art in the 1840s to the so-called “living substance” and “the physical basis of life” two decades later. I show that there were two major shifts in biological materialism that needed to occur before protoplasm theory could be elevated to have equal status with cell theory in the nineteenth century. First, I argue that biologists had to accept that life could inhere in matter alone, regardless of form. Second, I argue that in the 1840s, ideas of what formless, biological matter was capable of dramatically changed: going from a “coagulation paradigm” (Pickstone, 1973) that had existed since Theophrastus, to a more robust conception of matter that was itself capable of movement and self-maintenance. In addition to revisiting Schleiden and Schwann’s original writings on cell theory, this article looks especially closely at Hugo von Mohl’s definition of the protoplasm concept in 1846, how it differed from his primordial utricle theory of cell structure two years earlier. This article draws on Lakoff and Johnson’s theory of “ontological metaphors” to show that the cell, primordial utricle, and protoplasm can be understood as material container, object, and substance, and that these overlapping distinctions help explain the chaotic and confusing early history of cell theory.     

Object Segmentation from Motion Discontinuities and Temporal Occlusions–A Biologically Inspired Model

Background

Optic flow is an important cue for object detection. Humans are able to perceive objects in a scene using only kinetic boundaries, and can perform the task even when other shape cues are not provided. These kinetic boundaries are characterized by the presence of motion discontinuities in a local neighbourhood. In addition, temporal occlusions appear along the boundaries as the object in front covers the background and the objects that are spatially behind it.

Methodology/Principal Findings

From a technical point of view, the detection of motion boundaries for segmentation based on optic flow is a difficult task. This is due to the problem that flow detected along such boundaries is generally not reliable. We propose a model derived from mechanisms found in visual areas V1, MT, and MSTl of human and primate cortex that achieves robust detection along motion boundaries. It includes two separate mechanisms for both the detection of motion discontinuities and of occlusion regions based on how neurons respond to spatial and temporal contrast, respectively. The mechanisms are embedded in a biologically inspired architecture that integrates information of different model components of the visual processing due to feedback connections. In particular, mutual interactions between the detection of motion discontinuities and temporal occlusions allow a considerable improvement of the kinetic boundary detection.

Conclusions/Significance

A new model is proposed that uses optic flow cues to detect motion discontinuities and object occlusion. We suggest that by combining these results for motion discontinuities and object occlusion, object segmentation within the model can be improved. This idea could also be applied in other models for object segmentation. In addition, we discuss how this model is related to neurophysiological findings. The model was successfully tested both with artificial and real sequences including self and object motion.     

Short-Term Memory Maintenance of Object Locations during Active Navigation: Which Working Memory Subsystem Is Essential?

The goal of the present study was to examine the extent to which working memory supports the maintenance of object locations during active spatial navigation. Participants were required to navigate a virtual environment and to encode the location of a target object. In the subsequent maintenance period they performed one of three secondary tasks that were designed to selectively load visual, verbal or spatial working memory subsystems. Thereafter participants re-entered the environment and navigated back to the remembered location of the target. We found that while navigation performance in participants with high navigational ability was impaired only by the spatial secondary task, navigation performance in participants with poor navigational ability was impaired equally by spatial and verbal secondary tasks. The visual secondary task had no effect on navigation performance. Our results extend current knowledge by showing that the differential engagement of working memory subsystems is determined by navigational ability.     

How Can Selection of Biologically Inspired Features Improve the Performance of a Robust Object Recognition Model?

Humans can effectively and swiftly recognize objects in complex natural scenes. This outstanding ability has motivated many computational object recognition models. Most of these models try to emulate the behavior of this remarkable system. The human visual system hierarchically recognizes objects in several processing stages. Along these stages a set of features with increasing complexity is extracted by different parts of visual system. Elementary features like bars and edges are processed in earlier levels of visual pathway and as far as one goes upper in this pathway more complex features will be spotted. It is an important interrogation in the field of visual processing to see which features of an object are selected and represented by the visual cortex. To address this issue, we extended a hierarchical model, which is motivated by biology, for different object recognition tasks. In this model, a set of object parts, named patches, extracted in the intermediate stages. These object parts are used for training procedure in the model and have an important role in object recognition. These patches are selected indiscriminately from different positions of an image and this can lead to the extraction of non-discriminating patches which eventually may reduce the performance. In the proposed model we used an evolutionary algorithm approach to select a set of informative patches. Our reported results indicate that these patches are more informative than usual random patches. We demonstrate the strength of the proposed model on a range of object recognition tasks. The proposed model outperforms the original model in diverse object recognition tasks. It can be seen from the experiments that selected features are generally particular parts of target images. Our results suggest that selected features which are parts of target objects provide an efficient set for robust object recognition.     

Visuospatial Function in Early Alzheimer’s Disease—The Use of the Visual Object and Space Perception (VOSP) Battery

Alzheimer’s disease (AD) is the most frequent cause of dementia. The clinical symptoms of AD begin with impairment of memory and executive function followed by the gradual involvement of other functions, such as language, semantic knowledge, abstract thinking, attention, and visuospatial abilities. Visuospatial function involves the identification of a stimulus and its location and can be impaired at the beginning of AD. The Visual Object and Space Perception (VOSP) battery evaluates visuospatial function, while minimizing the interference of other cognitive functions.

Objectives

To evaluate visuospatial function in early AD patients using the VOSP and determine cutoff scores to differentiate between cognitively healthy individuals and AD patients.

Methods

Thirty-one patients with mild AD and forty-four healthy elderly were evaluated using a neuropsychological battery and the VOSP.

Results

In the VOSP, the AD patients performed more poorly in all subtests examining object perception and in two subtests examining space perception (Number Location and Cube Analysis). The VOSP showed good accuracy and good correlation with tests measuring visuospatial function.

Conclusion

Visuospatial function is impaired in the early stages of AD. The VOSP battery is a sensitive battery test for visuospatial deficits with minimal interference by other cognitive functions.     

What Are You or Who Are You? The Emergence of Social Interaction between Dog and an Unidentified Moving Object (UMO)

Robots offer new possibilities for investigating animal social behaviour. This method enhances controllability and reproducibility of experimental techniques, and it allows also the experimental separation of the effects of bodily appearance (embodiment) and behaviour. In the present study we examined dogs’ interactive behaviour in a problem solving task (in which the dog has no access to the food) with three different social partners, two of which were robots and the third a human behaving in a robot-like manner. The Mechanical UMO (Unidentified Moving Object) and the Mechanical Human differed only in their embodiment, but showed similar behaviour toward the dog. In contrast, the Social UMO was interactive, showed contingent responsiveness and goal-directed behaviour and moved along varied routes. The dogs showed shorter looking and touching duration, but increased gaze alternation toward the Mechanical Human than to the Mechanical UMO. This suggests that dogs’ interactive behaviour may have been affected by previous experience with typical humans. We found that dogs also looked longer and showed more gaze alternations between the food and the Social UMO compared to the Mechanical UMO. These results suggest that dogs form expectations about an unfamiliar moving object within a short period of time and they recognise some social aspects of UMOs’ behaviour. This is the first evidence that interactive behaviour of a robot is important for evoking dogs’ social responsiveness.     

The Effect of Short-Term Feeding of a High-Coconut Oil or High-Fat Diet on Neuroinflammation and the Performance of an Object–Place Task in Rats

The consumption of high-fat and high-sugar diets, in the form of junk food, and binge eating are now common. Increasing evidence suggests that a high-fat diet (HFD) can induce neuroinflammation and alter behavior. I aimed to study the effects of diets of differing fat content on neuroinflammation and spatial memory using an object–place (OP) task. Thirty-two adult male rats were allocated to four groups and fed a regular diet (Regular diet), a control diet (Control diet), an HFD (60% of calories from lard), or a high-coconut oil diet (HCOD; 60% of calories from coconut oil) for 3 days. Their water intake, food consumption, body mass, and metabolic variables were measured. HFD-fed rats showed significantly poorer performance on the OP task, as assessed using the discrimination index (??0.208?±?0.094), than the Regular (0.462?±?0.078; P?<?0.0001) and Control (0.379?±?0.081; P?=?0.0003) groups. However, no significant difference was observed in spatial memory between the HCOD and Regular groups. The concentrations of neuroinflammatory markers (interleukin [IL]-1β, IL-6, tumor necrosis factor α, and nuclear factor κB) were also measured in the hippocampus and prefrontal cortex. HFD-fed rats showed significantly higher levels of neuroinflammatory markers than the Regular and Control diet-fed groups. HCOD feeding did not induce neuroinflammation in the hippocampus and prefrontal cortex compared with the Regular and Control groups.

     

The “Bantu Clinic”: A Genealogy of the African Patient as Object and Effect of South African Clinical Medicine, 1930–1990

This paper is about power, medicine andthe identity of the African as a patient of westernmedicine. From a conventional perspective and asencoded in the current quest for wholeness thatcharacterises South African biomedical discourse, theAfrican patient – like any other patient – has alwaysexisted as an authentic and subjectified being, whosetrue attributes and experiences have been denied bythe mechanistic, reductionistic and ethnocentricpractices of clinical medicine. Against this liberalhumanist perspective on the body as ontologicallyindependent of power, this paper offers a Foucaultianreading of the African patient as – like any otherpatient – contingent upon the force relations immanentwithin and relayed through the clinical practices ofbiomedicine. A quintessential form of disciplinarymicro-power, these fabricate the most intimaterecesses of the human body as manageable objects ofmedical knowledge and social consciousness to makepossible the great control strategies of repression,segmentation and liberation that are the usual focusof conventional investigations into the place andfunction of medicine in society. Since the 1930s whenthe African body first emerged as a discrete object ofa secular clinical knowledge, these have repeatedlytransformed the attributes and identity of the Africanpatient, and the paper traces this archaeology ofSouth African clinical perception from then until the1990s to show how its quest for wholeness is not anend point of discovery or liberation, but merelyanother ephemeral crystallization of socio-medicalknowledge in a constantly changing force field ofdisciplinary power.     

Object recognition as a measure of memory in 1–2?years old transgenic minipigs carrying the APPsw mutation for Alzheimer’s disease

Alzheimer??s disease (AD) is a disabling, fatal disease, where animal models potentially can enable investigation of aetiology and treatment. The first litter of G?ttingen minipigs carrying a mutation for human AD was born in 2007, showing transgene expression. In human AD patients, memory impairment is the most striking and consistent feature. The aim of the present study was to examine effects of the APPsw transgene on memory of AD minipigs compared with non-transgenic controls at two ages (1?C2?years) using the spontaneous object recognition test (SORT), which is based on behavioural discrimination of familiar and novel objects. No significant difference between AD minipigs and controls was found when comparing object recognition as a measure of memory. The minipigs did explore the novel object significantly more than the familiar, indicating the expected recognition of the familiar object. Two different inter-phase intervals were used (IPI: 10?C40?min). For both ages, object recognition was evident using 10?min IPI. When using 40?min IPI, object recognition was evident only at age 1?year. Comparing memory of a relatively small group of AD minipigs and controls at two rather young ages using the SORT, we were not able to show memory impairment in APPsw carrying minipigs. Being an age-dependent disease, the transgene is expected to cause AD-like symptoms in this porcine model, and the SORT should be repeated at older ages.     

PDE5 Inhibition Improves Object Memory in Standard Housed Rats but Not in Rats Housed in an Enriched Environment: Implications for Memory Models?

Drug effects are usually evaluated in animals housed under maximally standardized conditions. However, it is assumed that an enriched environment (EE) more closely resembles human conditions as compared to maximally standardized laboratory conditions. In the present study, we examined the acute cognition enhancing effects of vardenafil, a PDE5 inhibitor, which stimulates protein kinase G/CREB signaling in cells, in three different groups of male Wistar rats tested in an object recognition task (ORT). Rats were either housed solitarily (SOL) or socially (SOC) under standard conditions, or socially in an EE. Although EE animals remembered object information longer in the vehicle condition, vardenafil only improved object memory in SOL and SOC animals. While EE animals had a heavier dorsal hippocampus, we found no differences between experimental groups in total cell numbers in the dentate gyrus, CA2–3 or CA1. Neither were there any differences in markers for pre- and postsynaptic density. No changes in PDE5 mRNA- and protein expression levels were observed. Basal pCREB levels were increased in EE rats only, whereas β-catenin was not affected, suggesting specific activation of the MAP kinase signaling pathway and not the AKT pathway. A possible explanation for the inefficacy of vardenafil could be that CREB signaling is already optimally stimulated in the hippocampus of EE rats. Since previous data has shown that acute PDE5 inhibition does not improve memory performance in humans, the use of EE animals could be considered as a more valid model for testing cognition enhancing drugs.     

Sex Differences in Object Manipulation in Wild Immature Chimpanzees (Pan troglodytes schweinfurthii) and Bonobos (Pan paniscus): Preparation for Tool Use?

Sex differences in immatures predict behavioural differences in adulthood in many mammal species. Because most studies have focused on sex differences in social interactions, little is known about possible sex differences in ‘preparation’ for adult life with regards to tool use skills. We investigated sex and age differences in object manipulation in immature apes. Chimpanzees use a variety of tools across numerous contexts, whereas bonobos use few tools and none in foraging. In both species, a female bias in adult tool use has been reported. We studied object manipulation in immature chimpanzees at Kalinzu (Uganda) and bonobos at Wamba (Democratic Republic of Congo). We tested predictions of the ‘preparation for tool use’ hypothesis. We confirmed that chimpanzees showed higher rates and more diverse types of object manipulation than bonobos. Against expectation, male chimpanzees showed higher object manipulation rates than females, whereas in bonobos no sex difference was found. However, object manipulation by male chimpanzees was play-dominated, whereas manipulation types of female chimpanzees were more diverse (e.g., bite, break, carry). Manipulation by young immatures of both species was similarly dominated by play, but only in chimpanzees did it become more diverse with age. Moreover, in chimpanzees, object types became more tool-like (i.e., sticks) with age, further suggesting preparation for tool use in adulthood. The male bias in object manipulation in immature chimpanzees, along with the late onset of tool-like object manipulation, indicates that not all (early) object manipulation (i.e., object play) in immatures prepares for subsistence tool use. Instead, given the similarity with gender differences in human children, object play may also function in motor skill practice for male-specific behaviours (e.g., dominance displays). In conclusion, even though immature behaviours almost certainly reflect preparation for adult roles, more detailed future work is needed to disentangle possible functions of object manipulation during development.     

<Emphasis Type="Italic">Umwelträume</Emphasis> and Multisensory Integration. Mirror Perspectives on the Subject–Object Dichotomy

This paper concerns epistemic developments in the field of sensory perception. I argue that Uexküll’s concept of the Umwelträume and certain principles of multisensory integration explain and describe in similar terms the manner in which different sensory modalities interact. Indeed, they both concern knowledge, describing in spatial terms how the mind makes itself up, makes up its objects, and how the objects, in turn, make up the mind. My intention is to set side by side these two trends of thought in order to mutually explain them. I suggest that there is some sort of dialectic at work between the Umwelträume and multisensory integration: in the first case interaction between the senses results in the creation of high definition subjects, whereas in the latter the interaction of the senses results in high definition objects.     

Object Depots in the Genus <Emphasis Type="Italic">Pogonomyrmex</Emphasis>: Exploring the “Who,” What,When, and Where

Harvester ants of the genus Pogonomyrmex collect and deposit many items on top of their nests. The depots of P. badius consist mostly of small charcoal fragments, while those of other species are primarily pebbles. Mature colonies can have hundreds of thousands of objects in their depot. In P. badius, the distributions of midden and charcoal about the mound are not completely overlapping, but are positively correlated in areas of overlap. Charcoal depots are isometric with colony size, but the amount of charcoal per colony size varies with season and site. The absence of the depot stimulates collection of nonfood objects. Recruitment to objects only occurs in the presence of food, and foragers choose objects based on size, but not color. An overview of current knowledge concerning depots in the genus indicates that depot formation is likely the ancestral state in the North American species. Furthermore, it is likely that selection is operating indirectly on these depots through selection on many dependent tasks.     

Specific Features of Morphology,Ecology, and Mitochondrial DNA Polymorphism of the Whitefish (<Emphasis Type="Italic">Coregonus lavaretus</Emphasis> L.) from the Keret’ River as a New Object for Artificial Propagation

The results of a study of the specific features of morphological and genetic polymorphism of the population of the anadromous White Sea common whitefish (Coregonus lavaretus L.) from the Keret’ River are presented. As assessed by its morphological features, the whitefish population of the Keret’ is homogenous. A comparison of the Keret’ whitefish with whitefish populations of both Europe and Siberia revealed clear differences. These differences relate first and foremost to the fish locomotion and orientation in space: body depth, length and positions of fins, and interorbital width. The characteristics of skull features (craniological analysis) of the White Sea whitefish are presented for the first time. The results of an analysis of the polymorphism of the ND1 fragment of the mitochondrial DNA confirmed an earlier suggestion about the mixed origin of the Keret’ River whitefish: the descendants of three different phylogenetic lineages were revealed in the population.