Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 2006. / Source: Dissertation Abstracts International, Volume: 67-07, Section: B, page: 4128. Adviser: Arthur F. Kramer. Includes bibliographical references (leaves 72-84) Available on microfilm from Pro Quest Information and Learning.

Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 2007. / Source: Dissertation Abstracts International, Volume: 69-02, Section: A, page: 0591. Advisers: Richard W. Sproat; Jennifer S. Cole. Includes bibliographical references (leaves 132-143) Available on microfilm from Pro Quest Information and Learning.

This thesis examines the relationship between folk psychology and scientific psychology, and argues that the conceptual taxonomy provided by the former is unsuitable for fine-grained cognitive scientific research. I avoid traditional eliminativism by reserving a role for folk psychology as a socio-normative discourse, where folk psychological concepts primarily refer to behaviour rather than to mental states, and also exert a regulative influence on behaviour. In the first half of this thesis I develop a positive account of folk psychology as a broad discourse that includes mental state attributions, behavioural predictions, narrative competency, and regulative mechanisms. In the second half I argue that the conceptual taxonomy provided by this discourse has led to theoretical confusions in both philosophy and cognitive science, and I propose a systematic methodology for developing a novel ‘cognitive ontology’ that is better suited for contemporary scientific research. What is folk psychology? In chapter 1 I survey the history of the term folk psychology and demonstrate that the term only really came into general usage following the work of Fodor and Churchland in the 1970s and 80s. I also argue that it is a mistake, stemming from this era, to identify folk psychology exclusively with propositional attitude psychology, which is just one particular way in which the folk might understand one another. If folk psychology is not just propositional attitude psychology, what else might it be? In chapter 2 I consider what I call the ‘universality assumption’, i.e. the assumption that folk psychological intuitions are shared across all cultures and languages. If this assumption were justified then it might provide partial support for the claim that folk psychology presents an accurate account of human cognition. However, there is significant evidence of variation in folk psychological intuitions, suggesting that folk psychology might be at least partially biased by cultural and linguistic influences. If folk psychology is not the same in every culture, how come it is so successful at predicting behaviour? In chapter 3 I look at various ways in which folk psychological discourse can play a regulative or normative role by exerting an influence on our behaviour. This role helps to explain how folk psychology can be predictively successful even if it fails to accurately describe the fine-grained details of human cognition, as via regulative mechanisms it is able to become a kind of self-fulfilling prophecy. How well does folk psychology match up with our scientific understanding of cognition? In chapter 4 I present evidence of cases where folk psychological concepts have served to mislead or confuse theoretical debates in philosophy of mind and cognitive science. I consider several case studies, including the false belief task in social cognition, the taxonomisation of sensory modalities, the extended cognition debate, and the recently emerging ‘Bayesian brain’ hypothesis. If folk psychological concepts do not refer to entities in our scientific theories, then what do they refer to? In chapter 5 I examine the status of folk psychological kinds as natural kinds, and argue that even under a very liberal account folk psychological kinds probably do not constitute viable scientific kinds. However, due to the regulative mechanisms described in chapter 3, they do constitute what Hacking has described as ‘human’ or ‘interactive’ kinds, which exhibit complex looping effects. What kinds of concepts should cognitive science use, if not folk psychological concepts? Finally, in chapter 6 I look at recent developments in ‘cognitive ontology’ revision and argue that we should adopt a systematic methodology for constructing novel concepts that better reflect our current best understanding of cognitive systems. In closing I consider the relationship between these novel concepts and the ontology presented by folk psychological discourse.

<p> The majority of computational theories of inductive processes in psychology derive from small-scale experiments with simple stimuli that are easy to represent. However, real-world stimuli are complex, hard to represent efficiently, and likely require very different cognitive strategies to cope with. Indeed, the difficulty of such tasks are part of what make humans so impressive, yet methodological resources for modeling their solutions are limited. This presents a fundamental challenge to the precision of psychology as a science, especially if traditional laboratory methods fail to generalize. Recently, a number of computationally tractable, data-driven methods such as deep neural networks have emerged in machine learning for deriving useful representations of complex perceptual stimuli, but they are explicitly optimized in service to engineering objectives rather than modeling human cognition. It has remained unclear to what extent engineering models, while often state-of-the-art in terms of human-level task performance, can be leveraged to model, predict, and understand humans.</p><p> In the following, I outline a methodology by which psychological research can confidently leverage representations learned by deep neural networks to model and predict complex human behavior, potentially extending the scope of the field. In Chapter 1, I discuss the challenges to ecological validity in the laboratory that may be partially circumvented by technological advances and trends in machine learning, and weigh the advantages and disadvantages of bootstrapping from largely uninterpretable models. In Chapter 2, I contrast methods from psychology and machine learning for representing complex stimuli like images. Chapter 3 provides a first case study of applying deep neural networks to predict whether objects in a large database of images will be remembered by humans. Chapter 4 provides the central argument for using representations from deep neural networks as proxies for human psychological representations in general. To do this, I establish and demonstrate methods for quantifying their correspondence, improving their correspondence with minimal cost, and applying the result to the modeling of downstream cognitive processes. Building on this, Chapter 5 develops a method for modeling human subjective probability over deep representations in order to capture multimodal mental visual concepts such as "landscape". Finally, in Chapter 6, I discuss the implications of the overall paradigm espoused in the current work, along with the most crucial challenges ahead and potential ways forward. The overall endeavor is almost certainly a stepping stone to methods that may look very different in the near future, as the gains in leveraging machine learning methods are consolidated and made more interpretable/useful. The hope is that a synergy can be formed between the two fields, each bootstrapping and learning from the other.</p><p>

<p> Visual perception can be understood as an inferential process that combines noisy sensory information with internalized knowledge drawn from previous experience. In statistical Bayesian terms, internal representations of the visual environment can be understood as posterior estimates obtained by weighting imperfect sensory information (a likelihood) by internalized biases (a prior). Given limited perceptual resources, it is advantageous for the visual system to capitalize on predictable regularities of the visual world, and internalize them in the form of priors. This dissertation presents novel findings in the domain of spatial vision and visual memory, as well as some new work on memory for the 3D orientation of objects. In all cases, an unprecedented signal-to-noise ratio, achieved by employing serial reproduction chains (a &ldquo;telephone game&rdquo; procedure) combined with non-parametric kernel density estimation techniques, reveals a number of stunning intricacies in the prior for the first time. Methodological implications, as well as implications for amending prior empirical findings and revisiting past theoretical explanations are discussed.</p><p>

<p> Visual search has been extensively studied in the laboratory, yielding broad insights into how we search through and attend to the world around us. In order to know if these insights are valid, however, this research must not be confined to the sanitized imagery typically found within the lab. Comparatively little research has been conducted on visual search within naturalistic settings, and this gap must therefore be bridged in order to further our understanding of visual search. Based on the results of Experiment 1, Experiment 2 was conducted to test three common effects observed in previous studies of visual search: the effects of background complexity, target-background similarity, and target-distractor similarity on response time. Results show that these hypotheses carry over to the natural world, but also that there are other effects present not accounted for by current theories of visual search. The argument is made for the modification of these theories to incorporate this naturalistic information. </p><p>

<p> The number of visual distraction-caused crashes highlights a need for non-visual information displays in vehicles. Auditory-supported air gesture controls could fill that need. This dissertation covers four experiments that aim to explore the design auditory-supported air gesture system and examine its real-world influence on driving performance. The first three experiments compared different prototype gesture control designs as participants used the systems in a driving simulator. The fourth experiment sought to answer more basic questions about how auditory displays influence performance in target acquisition tasks. Results from experiment 1 offered optimism for the potential of auditory-supported displays for navigating simple menus by showing a decrease in off-road glance time compared to visual-only displays. Experiment 1 also showed a need to keep menu items small in number but large in size. Results from experiment 2 showed auditory-supported air gesture controls can result in safer driving performance relative to touchscreens, but at the cost of slight decrements in menu task performance. Results from experiment 3 showed that drivers can navigate through simple menu structures totally eyes-free, with no visual displays, even with less effort compared to visual displays and visual plus auditory displays. Experiment 4 showed that auditory displays convey information and allow for accurate target selection, but result in slower selections and relatively less accurate selections compared to displays with visual information, especially for more difficult target selections. Overall, the experimental data highlight potential for auditory-supported air gesture controls for increasing eyes-on-road time relative to visual displays both in touchscreens and air gesture controls. However, this benefit came at a slight cost to target selection performance as participants generally took longer to process auditory information in simple target acquisition tasks. Experimental results are discussed in the context of multiple resource theory and Fitts&rsquo;s law. Design guidelines and future work are also discussed. </p><p>

<p> One conceptualization of emotional awareness is described as attention to one&rsquo;s emotions and clarity of one&rsquo;s emotions. Clarity has been further divided into source awareness (i.e., knowing the causes of emotions) and type awareness (identifying actual emotions). Emotional awareness has been examined in relation to suspiciousness, one of four cognitive-perceptual disturbances in schizotypy. Studies have not, however, examined all three facets of emotional awareness in the other three cognitive-perceptual disturbances and have not examined attributional styles in conjunction with emotional awareness and their relation to cognitive-perceptual disturbances in schizotypy. In addition, previous studies have not examined self-report measures of emotional awareness in conjunction with behavioral/qualitative measures. The current study examined these factors using a cross-sectional design. In this study, 178 undergraduates completed self-report measures of emotional awareness, cognitive-perceptual disturbances, emotional arousal, and attributional style, in addition to completing behavioral tasks assessing type awareness and attention to emotions and a qualitative interview assessing source awareness. Results showed that low type awareness significantly predicted ideas of reference (after controlling for emotional arousal, source awareness, and attention to emotions) and suspiciousness (after controlling for sex, emotional arousal, attention, and source awareness). In addition, low internality for negative events was significantly associated with suspiciousness and odd beliefs/magical thinking. These findings build upon previous work in this area and have implications for potential treatments for cognitive and perceptual disturbances associated with schizotypy. Future directions for additional research are also discussed. </p><p>

<p> The tendency for humans to give preferential attention to animate agents in their immediate surroundings has been well-documented and likely reflects an evolved specialization to a persistent adaptive problem. In uncertain or ambiguous cases, this tendency can result in an over-detection of animacy, as the potential costs of failing to detect an animate agent far outweigh those of mistaken identification. In line with this, it seems likely that humans have evolved a sensitivity to specific cues which are indicative of animacy such that the mere presence of these cues will lead to detection, regardless of the objective category membership of the entity in question. There exists a wealth of research speaking to this effect with regards to motion cues, specifically in terms of the capacity for self-propulsion and goal-directed action. Morphological cues have also been implicated - most especially the presence of facial features &ndash; as they specify a capacity for perceptual feedback from the environment, which is essential for goal-directed motion. However, it remains an open question as to whether the capacity for animacy detection is similarly sensitive to facial information in the absence of motion cues. </p><p> The experiments reported here attempted to address this question by implementing a novel task in which participants were asked to judge the animacy or inanimacy (or membership in animal or object categories) of different images: animals with and without visible facial features, and objects with and without visible facial features. Beyond replicating a general advantage for detecting animate agents over inanimate objects, the primary predictions for these experiments were that facial features would have a differential effect on performance, such that they would improve performance when visible in animals, and would hinder performance when visible in objects. Experiments 1a and 1b provided a preliminary confirmation of this pattern of responses using images of familiar and unfamiliar animals (e.g., dogs versus jellyfish), and unaltered images of objects with and without faces. Experiment 2 improved on the design of this task by more closely matching the sets of images (the same animals facing toward or away from the camera, and objects with faces which had been digitally altered to disrupt the facial features), and by changing the prompt of the task from yes/no judgments of animacy to categorization into animal or object groups. Experiment 3 examined the face inversion effect, or the failure to recognize familiar faces when their orientation is inverted, on animal-object categorization. Lastly, experiments 4 and 5 attempted to extend the findings from experiment 2 to preschool-aged children, by implementing a card sorting task (experiment 4) and a computerized animal detection task (experiment 5). The results of this series of experiments highlight the prominent role of facial features in detecting animate agents in one&rsquo;s surroundings.</p><p>

<p> Working memory is a collection of cognitive resources that allow for the temporary maintenance and manipulation of information. This information can then be used to accomplish task goals in a variety of different contexts. To do this, the working memory system is able to process many different kinds of information using resources dedicated to the processing of those specific types of information. This processing is modulated by a control component which is responsible for guiding actions in the face of interference. Recently, the way in which working memory handles the processing of this information has been the subject of debate. Specifically, current models of working memory differ in their conceptualization of its functional architecture and the interaction between domain-specific storage structures and domain-general control processes. Here, domain-specific processing is when certain components of a model are dedicated to processing certain kinds of information, be it spatial or verbal. Domain-general processing is a when a component of a model can process multiple kinds of information. One approach conceptualizes working memory as consisting of various discrete components that are dedicated to processing specific kinds of information. These multiple component models attempt to explain how domain-specific storage structures are coordinated by a domain-general control mechanism. They also predict that capacity variations in those domain-specific storage structures can directly affect the performance of the domain-general control mechanism. Another approach focuses primarily on the contributions of a domain-general control mechanism to behavior. These controlled attention approaches collapse working memory and attention and propose that a domain-general control mechanism is the primary source of individual differences. This means that variations in domain-specific storage structures are not predicted to affect the functioning of the domain-general control mechanism. This dissertation will make the argument that conceptualizing working memory as either domain-specific or domain-general creates a false dichotomy. To do this, different ways of measuring working memory capacity will first be discussed. That discussion will serve as a basis for understanding the differences, and similarities between both models. A more detailed exposition of both the multiple component model and controlled attention account will follow. Behavioral and physiological evidence will accompany the descriptions of both models. The emphasis of the evidence presented here will be on load effects: observed changes in task performance when information is maintained in working memory. Load effects can be specific to the type of information being maintained (domain-specific), or occur regardless of information type (domain-general). This dissertation will demonstrate how the two models fail to address evidence for both domain-specific and domain-general load effects. Given these inadequacies, a new set of experiments will be proposed that will seek to demonstrate both domain-specific and domain-general effects within the same paradigm. Being able to demonstrate both these effects will go some way towards accounting for the differing evidence presented in the literature. A brief conceptualization of a possible account to explain these effects will then be discussed. Finally, future directions for research will be described.</p><p>