No description available.

In military training contexts, fast and long term decisions are intermixed where survival and security are prioritized. Simulation-based training, here applied to ground patrols in Afghanistan, can provide preparation for mission critical and life critical operations prior to exposure to real danger. Optimising the effectiveness of simulation-based training raises the need for more detailed representations of the competences required, both for simulation design and for evaluating simulation effectiveness. These needs are here considered in terms of three research questions . The first research question asks how objects trigger dialogue in observational tasks. Eye gaze tracking and recorded dialogue provide a foundation for proposing the cognitive operational structures behind how objects and dialogue are structured when people work together when collaborating in simulation-based training sessions. The objects are tracked along with related observational tasks and the communication between people in a team in ground vehicles and in the Tactical Operations Centre (TOC). The second research question asks how the results of simulation-based training for emergency situations can be described and evaluated. The last research question asks how debriefing and learning create and refine cognitive comprehension, the competency developed in a group. Low level visual cognition in a tactical environment is explored using an eye gaze tracking system integrated with a simulation environment. The integrated system has been evaluated, its accuracy characterized, and the system was then used to evaluate hypotheses related to visual queuing and target selection. The research questions are then explored more broadly based upon two exploratory field studies of simulation-based training sessions held for military staff before leaving for ISAF in Afghanistan. Study methods here include eye gaze tracking, video and audio recording, behavioral observation and retrospective questions. The field studies were conducted at the Swedish Life Guard Regiment sub-departments: International Training Unit(IntUtbE), pre-deployment training for Peace support operations, and Swedish Armed Forces International Centre (SWEDINT), with their Simulation, Modeling and Practical Platform. Based upon data obtained in the field studies, cognitive models of decision processes involved in operational task performance are developed to provide a basis for answering the research questions. Cognitive modelling begins with the Belief, Desire and Intension (BDI) model. This model is then modified in several steps to cover different levels of decision making revealed by the field studies, including an intrapersonal and organizational layer, an educational layer, a layer where objects are build into the algorithm as a basis for purposive behavior, and finally a team competency layer built largely during debriefing sessions. These models can be used to evaluate simulation-based training effectiveness, to provide feedback both in real time and retrospectively to trainees and teams, and potentially could be used in operational systems to provide real-time information about individual and group state during operations, for decision enhancement, and potentially as elements of the implementation of automated operational forces.

Discovering a word’s part of speech is an essential step in acquiring the grammar of a language. In this thesis we examine a variety of computational Bayesian models that use linguistic input available to children, in the form of transcribed child directed speech, to learn part of speech categories. Part of speech categories are characterised by contextual (distributional/syntactic) and word-internal (morphological) similarity. In this thesis, we assume language learners will be aware of these types of cues, and investigate exactly how they can make use of them. Firstly, we enrich the context of a standard model (the Bayesian Hidden Markov Model) by adding sentence type to the wider distributional context.We show that children are exposed to a much more diverse set of sentence types than evident in standard corpora used for NLP tasks, and previous work suggests that they are aware of the differences between sentence type as signalled by prosody and pragmatics. Sentence type affects local context distributions, and as such can be informative when relying on local context for categorisation. Adding sentence types to the model improves performance, depending on how it is integrated into our models. We discuss how to incorporate novel features into the model structure we use in a flexible manner, and present a second model type that learns to use sentence type as a distinguishing cue only when it is informative. Secondly, we add a model of morphological segmentation to the part of speech categorisation model, in order to model joint learning of syntactic categories and morphology. These two tasks are closely linked: categorising words into syntactic categories is aided by morphological information, and finding morphological patterns in words is aided by knowing the syntactic categories of those words. In our joint model, we find improved performance vis-a-vis single-task baselines, but the nature of the improvement depends on the morphological typology of the language being modelled. This is the first token-based joint model of unsupervised morphology and part of speech category learning of which we are aware.

This thesis addresses the cognitive basis of syntactic adaptation, which biases speakers to repeat their own syntactic constructions and those of their conversational partners. I address two types of syntactic adaptation: short-term priming and longterm adaptation. I develop two metrics for syntactic adaptation within a speaker and between speakers in dialogue: one for short-term priming effects that decay quickly, and one for long-term adaptation over the course of a dialogue. Both methods estimate adaptation in large datasets consisting of transcribed human-human dialogue annotated with syntactic information. Two such corpora in English are used: Switchboard, a collection of spontaneous phone conversation, and HCRC Map Task, a set of task-oriented dialogues in which participants describe routes on a map to one another. I find both priming and long-term adaptation in both corpora, confirming well-known experimental results (e.g., Bock, 1986b). I extend prior work by showing that syntactic priming effects not only apply to selected syntactic constructions that are alternative realizations of the same semantics, but still hold when a broad variety of syntactic phrase structure rules are considered. Each rule represents a cognitive decision during syntactic processing. I show that the priming effect for a rule is inversely proportional to its frequency. With this methodology, I test predictions of the Interactive Alignment Model (IAM, Pickering and Garrod, 2004). The IAM claims that linguistic and situation model agreement between interlocutors in dialogue is the result of a cascade of resource-free, mechanistic priming effects on various linguistic levels. I examine task-oriented dialogue in Map Task, which provides a measure of task success through the deviance of the communicated routes on the maps. I find that long term syntactic adaptation predicts communicative success, and it does so earlier than lexical adaptation. The result is applied in a machine-learning based model that estimates task success based on the dialogue, capturing 14 percent of the variance in Map Task. Short-term syntactic priming differs qualitatively from long term adaptation, as it does not predict task success, providing evidence against learning as a single cognitive basis of adaptation effects. I obtain further evidence for the correlation between semantic activity and syntactic priming through a comparison of the Map Task and Switchboard corpora, showing that short-term priming is stronger in task-oriented dialogue than in spontaneous conversation. This difference is evident for priming between and within speakers, which suggests that priming is a mechanistic rather than strategic effect. I turn to an investigation of the level at which syntactic priming influences language production. I establish that the effect applies to structural syntactic decisions as opposed to all surface sequences of lexical categories. To do so, I identify pairs of part-of-speech categories which consistently cross constituent boundaries defined by the phrase structure analyses of the sentences. I show that such distituents are less sensitive to priming than pairs occurring within constituents. Thus, syntactic priming is sensitive to syntactic structure. The notion of constituent structure differs among syntactic models. Combinatory Categorial Grammar (CCG, Steedman, 2000) formalizes flexible constituent structure, accounting a varying degree of incrementality in syntactic sentence planning. I examine whether priming effects can support the predictions of CCG using the Switchboard corpus, which has been annotated with CCG syntax. I confirm the syntactic priming effect for lexical and non-lexical CCG categories, which encode partially satisfied subcategorization frames. I then show that both incremental and normal-form constituent structures exhibit priming, arguing for language production accounts that support flexible incrementality. The empirical results are reflected in a cognitive model of syntactic realization in language production. The model assumes that language production is subject to the same principles and constraints as any other form of cognition and follows the ACT-R framework (Anderson et al., 2004). Its syntactic process implements my empirical results on priming and is based on CCG. Syntactic planning can take place incrementally and non-incrementally. The model is able to generate simple sentences that vary syntactically, similar to the materials used in the experimental priming literature. Syntactic adaptation emerges due to a preferential and sped-up memory retrieval of syntactic categories describing linearization and subcategorization requirements. Long-term adaptation is explained as a form of learning, while shortterm priming is the result of a combination of learning and spreading activation from semantic and lexical material. Simulations show that the model produces the adaptation effects and their inverse frequency interaction, as well as cumulativity of long-term adaptation.

An investigation is conducted into the cognitive effects of using different computer based instructions media in acquisition of specific novel human skills. With recent rapid advances in computing and multimedia instructional delivery, several contemporary research have focussed on the best practices for training and learning delivered via computer based multimedia simulations. More often than not, the aim has been cost minimisation through an optimisation of the instructional delivery process for efficient knowledge acquisition. The outcome of such research effort in general have been largely divergent and inconclusive. The work reported in this thesis utilises a dual prong methodology to provide a novel perspective on the moderating effects of computer based instructional visualisations with a focus on the interaction of interface dynamism with target knowledge domains and trainee cognitive characteristics. The first part of the methodology involves a series of empirical experiments that incrementally measures/compares the cognitive benefits of different levels of instructional interface dynamism for efficient task representation and post-acquisition skilled performance. The first of these experiments utilised a mechanical disassembly task to investigate novel acquisition of procedural motor skills by comparing task comprehension and performance. The other experiments expanded the initial findings to other knowledge domains as well as controlled for potential confounding variables. The integral outcome of these experiments helped to define a novel framework for describing multimodal perception of different computer based instruction types and its moderating effect on post-learning task performance. A parallel computational cognitive modelling effort provided the complementary methodology to investigate cognitive processing associated with different instructional interfaces at a lower level of detail than possible through empirical observations. Novel circumventions of some existing limitations of the selected ACT-R 6.0 cognitive modelling architecture were proposed to achieve the precision required. The ACT-R modifications afforded the representation of human motor movements at an atomic level of detail and with a constant velocity profile as opposed to what is possible with the default manual module. Additional extensions to ACT-R 6.0 also allowed accurate representation of the noise inherent in the recall of spatial locations from declarative memory. The method used for this representation is potentially extendable for application to 3-D spatial representation in ACT-R. These novel propositions are piloted in a proof-of-concept effort followed by application to a more complete, naturally occurring task sequence. The modelling methodology is validated with established human data of skilled task performances. The combination of empirical observations and detailed cognitive modelling afforded novel insights to the hitherto controversial findings on the cognitive benefits of different multimodal instructional presentations. The outcome has implications for training research and development involving computer based simulations.

The requirements for virtual agents are more and more demanding. In order to manage the complex behavior of the agent, it's possible to take advantage of cognitive architectures which arised on the field neuroscience and artificial intelligence. This work examines PoJACTR library which links Pogamut library for developing intelligent agents in Unreal Tournament 2004 and jACT-R library which is Java implementation of one of the leading cognitive architectures ACTR. This work also studies certain agent implementation problems in PoJACTR and proposes a solution for them in form of debugging tools, which were subsequently implemented on an Eclipse IDE platform. In addition, it expands PoJACTR navigation and communication library modules for the game - Capture The Flag. As a validation, two agents (bots) were developed to play game, one in standard Pogamut and one in PoJACTR. When matched against each other in battle, Po- JACTR bot had comparable performance to a Pogamut bot. The results showed that debugging tools facilitated development process of PoJACTR agents.

The requirements for virtual agents are more and more demanding. In order to manage the complex behavior of the agent, it's possible to take advantage of cognitive architectures which arised on the field neuroscience and artificial intelligence. This work examines PoJACTR library which links Pogamut library for developing intelligent agents in Unreal Tournament 2004 and jACT-R library which is Java implementation of one of the leading cognitive architectures ACTR. This work also studies certain agent implementation problems in PoJACTR and proposes a solution for them in form of debugging tools, which were subsequently implemented on an Eclipse IDE platform. In addition, it expands PoJACTR navigation and communication library modules for the game - Capture The Flag. As a validation, two agents (bots) were developed to play game, one in standard Pogamut and one in PoJACTR. When matched against each other in battle, PoJACTR bot had comparable performance to a Pogamut bot. The results showed that debugging tools facilitated development process of PoJACTR agents. Powered by TCPDF (www.tcpdf.org)

Cultural artifacts, such as language, survive and replicate by passing from mind to mind. Cultural evolution always proceeds by an inductive process, where behaviors are never directly copied, but reverse engineered by the cognitive mechanisms involved in learning and production. I will refer to this type of evolutionary change as inductive evolution and explain how this represents a broader class of evolutionary processes that can include both neutral and selective evolution. This thesis takes a mechanistic approach to understanding the forces of evolution underlying change in culture over time, where the mechanisms of change are sought within human cognition. I define culture as anything that replicates by passing through a cognitive system and take language as a premier example of culture, because of the wealth of knowledge about linguistic behaviors (external language) and its cognitive processing mechanisms (internal language). Mainstream cultural evolution theories related to social learning and social transmission of information define culture ideationally, as the subset of socially-acquired information in cognition that affects behaviors. Their goal is to explain behaviors with culture and avoid circularity by defining behaviors as markedly not part of culture. I take a reductionistic approach and argue that all there is to culture is brain states and behaviors, and further, that a complete explanation of the forces of cultural change can not be explained by a subset of cognition related to social learning, but necessarily involves domain-general mechanisms, because cognition is an integrated system. Such an approach should decompose culture into its constituent parts and explore 1) how brains states effect behavior, 2) how behavior effects brain states, and 3) how brain states and behaviors change over time when they are linked up in a process of cultural transmission, where one person's behavior is the input to another. I conduct several psychological experiments on frequency learning with adult learners and describe the behavioral biases that alter the frequencies of linguistic variants over time. I also fit probabilistic models of cognition to participant data to understand the inductive biases at play during linguistic frequency learning. Using these inductive and behavioral biases, I infer a Markov model over my empirical data to extrapolate participants' behavior forward in cultural evolutionary time and determine equivalences (and divergences) between inductive evolution and standard models from population genetics. As a key divergence point, I introduce the concept of non-binomial cultural drift, argue that this is a rampant form of neutral evolution in culture, and empirically demonstrate that probability matching is one such inductive mechanism that results in non-binomial cultural drift. I argue further that all inductive problems involving representativeness are potential drivers of neutral evolution unique to cultural systems. I also explore deviations from probability matching and describe non-neutral evolution due to inductive regularization biases in a linguistic and non-linguistic domain. Here, I offer a new take on an old debate about the domain-specificity vs -generality of the cognitive mechanisms involved in language processing, and show that the evolution of regularity in language cannot be predicted in isolation from the general cognitive mechanisms involved in frequency learning. Using my empirical data on regularization vs probability matching, I demonstrate how the use of appropriate non-binomial null hypotheses offers us greater precision in determining the strength of selective forces in cultural evolution.

The aim of Human Computer Interaction or HCI is to both understand and improve the quality of the users' experience with the systems and technology they interact with. Recent HCI research requirements have stated a need for a unified predictive approach to system design that consolidates system engineering, cognitive modelling, and design principles into a single 'total system approach.' At present, few methods seek to integrate all three of these aspects into a single method and of those that do many are extensions to existing engineering techniques. This thesis, however proposes a new behaviour based approach designed to identify usability problems early in the design process before testing the system with actual users. In order to address the research requirements, this model uses a new design notation called Genetic Software Engineering (GSE) in conjunction with aspects of a cognitive modelling technique called NGOMSL (Natural GOMS Language) as the basis for this approach. GSE's behaviour tree notation, and NGOMSL's goal orientated format are integrated using a set of simple conversion rules defined in this study. Several well established design principles, believed to contribute to the eventual usability of a product, are then modelled in GSE. This thesis addresses the design of simple interfaces and the design of complex ubiquitous technology. The new GSE approach is used to model and predict usability problems in an extensive range of tasks from programming a VCR to making a video recording on a modern mobile phone. The validity of these findings is tested against actual user tests on the same tasks and devices to demonstrate the effectiveness of the GSE approach. Ultimately, the aim of the study is to demonstrate the effectiveness of the new cognitive and engineering based approach at predicting usability problems based on tangible representations of established design principles. This both fulfils the MCI research requirements for a 'total system approach' and establishes a new and novel approach to user interface and system design.

An issue widely discussed in robotics research is the ageing society with its consequences for care-giving institutions and opportunities for developments in the area of service robots and robot companions. The general idea of using robotic systems in a personal or private context to support an independent way of living not only for the elderly but also for the physically impaired is pursued in different ways, ranging from socially oriented robotic pets to mobile assistants. Thus, the idea of the personalised general service robot is not too far fetched. Crucial for such a service robot is the ability to navigate in its working environment, which has to be assumed an arbitrary domestic or office-like environment that is shared with human users and bystanders. With methods developed and investigated in the field of simultaneous localisation and mapping it has become possible for mobile robots to explore and map an unknown environment, while they can stay localised with respect to their starting point and the surroundings. These approaches though do not consider the representation of the environment that is used by humans to refer to particular places. Robotic maps are often metric representations of features that can be obtained from sensory data. Humans have a more topological, in fact partially hierarchical way of representing environments. Especially for the communication between a user and her personal robot it is thus necessary to provide a link between the robotic map and the human understanding of the robot's workspace. The term Human Augmented Mapping is used for a framework that allows to integrate a robotic map with human concepts. Communication about the environment can thus be facilitated. By assuming an interactive setting for the map acquisition process it is possible for the user to influence the process significantly. Personal preferences can be made part of the environment representation that is acquired by the robot. Advantages become also obvious for the mapping process itself, since in an interactive setting the robot can ask for information and resolve ambiguities with the help of the user. Thus, a scenario of a ``guided tour'' in which a user can ask a robot to follow and present the surroundings is assumed as the starting point for a system for the integration of robotic mapping, interaction and human environment representations. A central point is the development of a generic, partially hierarchical environment model, that is applied in a topological graph structure as part of an overall experimental Human Augmented Mapping system implementation. Different aspects regarding the representation of entities of the spatial concepts used in this hierarchical model, particularly considering regions, are investigated. The proposed representation is evaluated both as description of delimited regions and for the detection of transitions between them. In three user studies different aspects of the human-robot interaction issues of Human Augmented Mapping are investigated and discussed. Results from the studies support the proposed model and representation approaches and can serve as basis for further studies in this area. / QC 20100914