A persistent issue of debate in the area of 3D object recognition concerns the nature of the experientially acquired object models in the primate visual system. One prominent proposal in this regard has expounded the use of object centered models, such as representations of the objects' 3D structures in a coordinate frame independent of the viewing parameters [Marr and Nishihara, 1978]. In contrast to this is another proposal which suggests that the viewing parameters encountered during the learning phase might be inextricably linked to subsequent performance on a recognition task [Tarr and Pinker, 1989; Poggio and Edelman, 1990]. The 'object model', according to this idea, is simply a collection of the sample views encountered during training. Given that object centered recognition strategies have the attractive feature of leading to viewpoint independence, they have garnered much of the research effort in the field of computational vision. Furthermore, since human recognition performance seems remarkably robust in the face of imaging variations [Ellis et al., 1989], it has often been implicitly assumed that the visual system employs an object centered strategy. In the present study we examine this assumption more closely. Our experimental results with a class of novel 3D structures strongly suggest the use of a view-based strategy by the human visual system even when it has the opportunity of constructing and using object-centered models. In fact, for our chosen class of objects, the results seem to support a stronger claim: 3D object recognition is 2D view-based.

We address the problem of fast relabeling of deformable Delaunay tetrahedral meshes using a compact uniform grid, with CPU parallelization through OpenMP. This problem is important in visualizing the simulation of deformable objects and arises in scientific visualization, games, computer vision, and motion picture production. Many existing software tools and APIs have the ability to manipulate 3D virtual objects. Prior mesh-based representations either allow topology changes or are fast. We aim for both. Specifically, we improve the efficiency of the relabeling step in the Delaunay deformable mesh invented by Pons and Boissonnat and improved by Tychonievich and Jones. The relabeling step assigns material types to deformed meshes and accounts for 70% of the computation time of Tychonievich and Jones' algorithm. We have designed a deformable mesh algorithm using a Delaunay triangulation and a compact uniform grid with CPU parallelization to obtain greater speed than other methods that support topology changes. On average, over all our experiments and with various 3D objects, the serial implementation of the relabeling step of our work reports a speedup of 2.145 over the previous fastest method, including one outlier whose speedup was 3.934. When running in parallel on 4 cores, on average the relabeling step of our work achieves a speedup of 3.979, with an outlier at 7.63. The average speedup of our parallel relabeling step over our own serial relabeling step is 1.841.Simulation results show that the resulting mesh supports topology changes.

This thesis project had the overall aim of investigating older adults (OA) representation of real-world objects through different cognitive paradigms probing visual working memory (vWM) and attention. OAs are known to demonstrate difficulties in these two cognitive domains, however, an under-researched in the aging field is the possible influence of the quality of representations in conjunction with the amount of representations in determining OA’s performances. Moreover, a majority of research in the aging field has utilized geometrical shapes, therefore I was interested to investigate whether the previous findings extend when real-world objects are used. Accordingly, the main questions accordingly probed in this study were 1) whether the content of object representations influence OAs’ performances when vWM and attention are probed, or whether effects are solely due to the amount, and 2) whether age-effects found in previous vWM and attention studies in aging, extend when real-world objects are used. To this aim, Chapter 3 describes two experiments where I manipulated the perceptual and conceptual similarity amongst objects presented to OAs and young adults (YAs) when performing a change detection task taxing vWM. Results indicated both age groups had significantly lower performances when there 3 objects to retain and recall, but no age-related differences in the maintenance and recall of these types of objects. Chapter 4 and Chapter 5 report three experiments addressing the impact of object numerosity and/or nature in OAs’ attention deployment during a visual search task. Results of two experiments in Chapter 4 with manual responses found that OAs’ performance was detrimentally impacted by the amount of object representations, irrespective of the relation amongst the items. Chapter 5 illustrates an eye movement study, aimed at exploring more in detail the visual exploratory behavior of older and young individuals. Overall, OAs were slower and showed a larger tendency to get distracted than YAs, regardless of the perceptual or conceptual target-distractor similarity. Overall, the studies presented in this thesis indicated that aging negatively affects the ability to ignore distracting information. However, little evidence was provided as to whether aging negatively influences the content representation of an object. This result confirms previous research indicating that some, but not all aspects, decline in late adulthood.

Micro-affordance effects have been reported for several different components of the reach-to-grasp action during both on-line and off-line visual processing. The presence of such effects represents a strong demonstration of the close relationship between perception, action, and cognition. In this thesis 7 experiments are described, which investigate different aspects of that relationship, with particular attention on the nature of object representations. In 5 behavioural experiments as well as in 1 Transcranial Magnetic Stimulation (TMS) experiment a stimulus-response compatibility paradigm is employed to examine the presence of micro-affordance effects arising during language processing of object names. The power and precision component of the reach-to-grasp action is investigated in relation to the compatibility of an object for grasping with either a power or a precision grasp. Overall, the results of the experiments discussed in the present thesis suggest that: a) object representations activated during language processing of object names are able to potentiate actions arising from the component of the reach-to-grasp action under investigation; b) such representations might be more semantic or „propositional‟ than depictive in nature, therefore more related to stored semantic knowledge of the object and its associated actions than to its detailed visual properties; c) this semantic information about objects seems to be automatically translated into specific motor activity, even in the absence of any intention to act; d) finally, such semantic, non-visual motor potentiation seems to be rapid and relatively short lived.

<p>This thesis demonstrates that infants represent temporarily non-visible, or occluded, objects. From 4 months of age, infants could accurately predict the reappearance of a moving object after 660 ms of non visibility; indicating accurate spatiotemporal representations. At this age predictions were dominated by associations between specific events and outcomes (associative rules). Between 6 and 8 months of age predictions became dominated by extrapolations (Study III). From 6 months infants could represent occluded objects for up to 4 seconds. The number of successful predictions decreased, however, if the information contained in the occlusion event diminished (time of accretion and deletion). As infants grew older (up to 12 months) they produced more accurate predictions. (Study II). The similarities between adult and infant performances were numerous (Study I). These conclusion are based on one cross sectional (Study I) and two longitudinal studies (Study II & III) in which an object, a ‘happy face’, moved on circular (Study I, II, & III) and other complex trajectories (Study III). One portion of each trajectory was covered by a screen that blocked the object from sight. In each study participants gaze were recorded with an infrared eye tracking system (ASL 504) and a magnetic head tracker (Flock of Birds). This data was combined with data from the stimulus and stored for of line analysis.</p>

This thesis demonstrates that infants represent temporarily non-visible, or occluded, objects. From 4 months of age, infants could accurately predict the reappearance of a moving object after 660 ms of non visibility; indicating accurate spatiotemporal representations. At this age predictions were dominated by associations between specific events and outcomes (associative rules). Between 6 and 8 months of age predictions became dominated by extrapolations (Study III). From 6 months infants could represent occluded objects for up to 4 seconds. The number of successful predictions decreased, however, if the information contained in the occlusion event diminished (time of accretion and deletion). As infants grew older (up to 12 months) they produced more accurate predictions. (Study II). The similarities between adult and infant performances were numerous (Study I). These conclusion are based on one cross sectional (Study I) and two longitudinal studies (Study II &amp; III) in which an object, a ‘happy face’, moved on circular (Study I, II, &amp; III) and other complex trajectories (Study III). One portion of each trajectory was covered by a screen that blocked the object from sight. In each study participants gaze were recorded with an infrared eye tracking system (ASL 504) and a magnetic head tracker (Flock of Birds). This data was combined with data from the stimulus and stored for of line analysis.