ARTSCENE: A Neural System for Natural Scene Classification

How do humans rapidly recognize a scene? How can neural models capture this biological competence to achieve state-of-the-art scene classification? The ARTSCENE neural system classifies natural scene photographs by using multiple spatial scales to efficiently accumulate evidence for gist and texture. ARTSCENE embodies a coarse-to-fine Texture Size Ranking Principle whereby spatial attention processes multiple scales of scenic information, ranging from global gist to local properties of textures. The model can incrementally learn and predict scene identity by gist information alone and can improve performance through selective attention to scenic textures of progressively smaller size. ARTSCENE discriminates 4 landscape scene categories (coast, forest, mountain and countryside) with up to 91.58% correct on a test set, outperforms alternative models in the literature which use biologically implausible computations, and outperforms component systems that use either gist or texture information alone. Model simulations also show that adjacent textures form higher-order features that are also informative for scene recognition.

Dopaminergic and Non-Dopaminergic Value Systems in Conditioning and Outcome-Specific Revaluation

Animals are motivated to choose environmental options that can best satisfy current needs. To explain such choices, this paper introduces the MOTIVATOR (Matching Objects To Internal Values Triggers Option Revaluations) neural model. MOTIVATOR describes cognitiveemotional interactions between higher-order sensory cortices and an evaluative neuraxis composed of the hypothalamus, amygdala, and orbitofrontal cortex. Given a conditioned stimulus (CS), the model amygdala and lateral hypothalamus interact to calculate the expected current value of the subjective outcome that the CS predicts, constrained by the current state of deprivation or satiation. The amygdala relays the expected value information to orbitofrontal cells that receive inputs from anterior inferotemporal cells, and medial orbitofrontal cells that receive inputs from rhinal cortex. The activations of these orbitofrontal cells code the subjective values of objects. These values guide behavioral choices. The model basal ganglia detect errors in CS-specific predictions of the value and timing of rewards. Excitatory inputs from the pedunculopontine nucleus interact with timed inhibitory inputs from model striosomes in the ventral striatum to regulate dopamine burst and dip responses from cells in the substantia nigra pars compacta and ventral tegmental area. Learning in cortical and striatal regions is strongly modulated by dopamine. The model is used to address tasks that examine food-specific satiety, Pavlovian conditioning, reinforcer devaluation, and simultaneous visual discrimination. Model simulations successfully reproduce discharge dynamics of known cell types, including signals that predict saccadic reaction times and CS-dependent changes in systolic blood pressure.

Spikes, Synchrony, and Attentive Learning by Laminar Thalamocortical Circuits

This article develops the Synchronous Matching Adaptive Resonance Theory (SMART) neural model to explain how the brain may coordinate multiple levels of thalamocortical and corticocortical processing to rapidly learn, and stably remember, important information about a changing world. The model clarifies how bottom-up and top-down processes work together to realize this goal, notably how processes of learning, expectation, attention, resonance, and synchrony are coordinated. The model hereby clarifies, for the first time, how the following levels of brain organization coexist to realize cognitive processing properties that regulate fast learning and stable memory of brain representations: single cell properties, such as spiking dynamics, spike-timing-dependent plasticity (STDP), and acetylcholine modulation; detailed laminar thalamic and cortical circuit designs and their interactions; aggregate cell recordings, such as current-source densities and local field potentials; and single cell and large-scale inter-areal oscillations in the gamma and beta frequency domains. In particular, the model predicts how laminar circuits of multiple cortical areas interact with primary and higher-order specific thalamic nuclei and nonspecific thalamic nuclei to carry out attentive visual learning and information processing. The model simulates how synchronization of neuronal spiking occurs within and across brain regions, and triggers STDP. Matches between bottom-up adaptively filtered input patterns and learned top-down expectations cause gamma oscillations that support attention, resonance, and learning. Mismatches inhibit learning while causing beta oscillations during reset and hypothesis testing operations that are initiated in the deeper cortical layers. The generality of learned recognition codes is controlled by a vigilance process mediated by acetylcholine.

Towards a Unified Theory of Neocortex: Laminar Cortical Circuits for Vision and Cognition

A key goal of computational neuroscience is to link brain mechanisms to behavioral functions. The present article describes recent progress towards explaining how laminar neocortical circuits give rise to biological intelligence. These circuits embody two new and revolutionary computational paradigms: Complementary Computing and Laminar Computing. Circuit properties include a novel synthesis of feedforward and feedback processing, of digital and analog processing, and of pre-attentive and attentive processing. This synthesis clarifies the appeal of Bayesian approaches but has a far greater predictive range that naturally extends to self-organizing processes. Examples from vision and cognition are summarized. A LAMINART architecture unifies properties of visual development, learning, perceptual grouping, attention, and 3D vision. A key modeling theme is that the mechanisms which enable development and learning to occur in a stable way imply properties of adult behavior. It is noted how higher-order attentional constraints can influence multiple cortical regions, and how spatial and object attention work together to learn view-invariant object categories. In particular, a form-fitting spatial attentional shroud can allow an emerging view-invariant object category to remain active while multiple view categories are associated with it during sequences of saccadic eye movements. Finally, the chapter summarizes recent work on the LIST PARSE model of cognitive information processing by the laminar circuits of prefrontal cortex. LIST PARSE models the short-term storage of event sequences in working memory, their unitization through learning into sequence, or list, chunks, and their read-out in planned sequential performance that is under volitional control. LIST PARSE provides a laminar embodiment of Item and Order working memories, also called Competitive Queuing models, that have been supported by both psychophysical and neurobiological data. These examples show how variations of a common laminar cortical design can embody properties of visual and cognitive intelligence that seem, at least on the surface, to be mechanistically unrelated.

The Synthesis of Arbitrary Stable Dynamics in Non-linear Neural Networks II: Feedback and Universality

We wish to construct a realization theory of stable neural networks and use this theory to model the variety of stable dynamics apparent in natural data. Such a theory should have numerous applications to constructing specific artificial neural networks with desired dynamical behavior. The networks used in this theory should have well understood dynamics yet be as diverse as possible to capture natural diversity. In this article, I describe a parameterized family of higher order, gradient-like neural networks which have known arbitrary equilibria with unstable manifolds of known specified dimension. Moreover, any system with hyperbolic dynamics is conjugate to one of these systems in a neighborhood of the equilibrium points. Prior work on how to synthesize attractors using dynamical systems theory, optimization, or direct parametric. fits to known stable systems, is either non-constructive, lacks generality, or has unspecified attracting equilibria. More specifically, We construct a parameterized family of gradient-like neural networks with a simple feedback rule which will generate equilibrium points with a set of unstable manifolds of specified dimension. Strict Lyapunov functions and nested periodic orbits are obtained for these systems and used as a method of synthesis to generate a large family of systems with the same local dynamics. This work is applied to show how one can interpolate finite sets of data, on nested periodic orbits.

Reasoning about trust, entrepreneurial services, and the virtual team

This Portfolio of Exploration (PoE) tracks a transformative learning developmental journey that is directed at changing meaning making structures and mental models within an innovation practice. The explicit purpose of the Portfolio is to develop new and different perspectives that enable the handling of new and more complex phenomena through self transformation and increased emotional intelligence development. The Portfolio provides a response to the question: ‘What are the key determinants that enable a Virtual Team (VT) to flourish where flourishing means developing and delivering on the firm’s innovative imperatives?’ Furthermore, the PoE is structured as an investigation into how higher order meaning making promotes ‘entrepreneurial services’ within an intra-firm virtual team, with a secondary aim to identify how reasoning about trust influence KGPs to exchange knowledge. I have developed a framework which specifically focuses on the effectiveness of any firms’ Virtual Team (VT) through transforming the meaning making of the VT participants. I hypothesized it is the way KGPs make meaning (reasoning about trust) which differentiates the firm as a growing firm in the sense of Penrosean resources: ‘inducement to expand and a limit of expansion’ (1959). Reasoning about trust is used as a higher order meaning-making concept in line with Kegan’s (1994) conception of complex meaning making, which is the combining of ideas and data in ways that transform meaning and implicates participants to find new ways of knowledge generation. Simply, it is the VT participants who develop higher order meaning making that hold the capabilities to transform the firm from within, providing a unique competitive advantage that enables the firm to grow.

Development and characterization of novel transgenic techniques for the study of neuronal circuitry

Modern neuroscience relies heavily on sophisticated tools that allow us to visualize and manipulate cells with precise spatial and temporal control. Transgenic mouse models, for example, can be used to manipulate cellular activity in order to draw conclusions about the molecular events responsible for the development, maintenance and refinement of healthy and/or diseased neuronal circuits. Although it is fairly well established that circuits respond to activity-dependent competition between neurons, we have yet to understand either the mechanisms underlying these events or the higher-order plasticity that synchronizes entire circuits. In this thesis we aimed to develop and characterize transgenic mouse models that can be used to directly address these outstanding biological questions in different ways. We present SLICK-H, a Cre-expressing mouse line that can achieve drug-inducible, widespread, neuron-specific manipulations in vivo. This model is a clear improvement over existing models because of its particularly strong, widespread, and even distribution pattern that can be tightly controlled in the absence of drug induction. We also present SLICK-V::Ptox, a mouse line that, through expression of the tetanus toxin light chain, allows long-term inhibition of neurotransmission in a small subset (<1%) of fluorescently labeled pyramidal cells. This model, which can be used to study how a silenced cell performs in a wildtype environment, greatly facilitates the in vivo study of activity-dependent competition in the mammalian brain. As an initial application we used this model to show that tetanus toxin-expressing CA1 neurons experience a 15% - 19% decrease in apical dendritic spine density. Finally, we also describe the attempt to create additional Cre-driven mouse lines that would allow conditional alteration of neuronal activity either by hyperpolarization or inhibition of neurotransmission. Overall, the models characterized in this thesis expand upon the wealth of tools available that aim to dissect neuronal circuitry by genetically manipulating neurons in vivo.

Learning to drive: from hazard detection to hazard handling

Hazard perception has been found to correlate with crash involvement, and has thus been suggested as the most likely source of any skill gap between novice and experienced drivers. The most commonly used method for measuring hazard perception is to evaluate the perception-reaction time to filmed traffic events. It can be argued that this method lacks ecological validity and may be of limited value in predicting the actions drivers’ will take to hazards encountered. The first two studies of this thesis compare novice and experienced drivers’ performance on a hazard detection test, requiring discrete button press responses, with their behaviour in a more dynamic driving environment, requiring hazard handling ability. Results indicate that the hazard handling test is more successful at identifying experience-related differences in response time to hazards. Hazard detection test scores were strongly related to performance on a driver theory test, implying that traditional hazard perception tests may be focusing more on declarative knowledge of driving than on the procedural knowledge required to successfully avoid hazards while driving. One in five Irish drivers crash within a year of passing their driving test. This suggests that the current driver training system does not fully prepare drivers for the dangers they will encounter. Thus, the third and fourth studies in this thesis focus on the development of two simulator-based training regimes. In the third study participants receive intensive training on the molar elements of driving i.e. speed and distance evaluation. The fourth study focuses on training higher order situation awareness skills, including perception, comprehension and projection. Results indicate significant improvement in aspects of speed, distance and situation awareness across training days. However, neither training programme leads to significant improvements in hazard handling performance, highlighting the difficulties of applying learning to situations not previously encountered.

Nonlinear piezoelectricity in electroelastic energy harvesters: Modeling and experimental identification

We propose and experimentally validate a first-principles based model for the nonlinear piezoelectric response of an electroelastic energy harvester. The analysis herein highlights the importance of modeling inherent piezoelectric nonlinearities that are not limited to higher order elastic effects but also include nonlinear coupling to a power harvesting circuit. Furthermore, a nonlinear damping mechanism is shown to accurately restrict the amplitude and bandwidth of the frequency response. The linear piezoelectric modeling framework widely accepted for theoretical investigations is demonstrated to be a weak presumption for near-resonant excitation amplitudes as low as 0.5 g in a prefabricated bimorph whose oscillation amplitudes remain geometrically linear for the full range of experimental tests performed (never exceeding 0.25% of the cantilever overhang length). Nonlinear coefficients are identified via a nonlinear least-squares optimization algorithm that utilizes an approximate analytic solution obtained by the method of harmonic balance. For lead zirconate titanate (PZT-5H), we obtained a fourth order elastic tensor component of c1111p =-3.6673× 1017 N/m2 and a fourth order electroelastic tensor value of e3111 =1.7212× 108 m/V. © 2010 American Institute of Physics.

Histone modifications within the human X centromere region.

Human centromeres are multi-megabase regions of highly ordered arrays of alpha satellite DNA that are separated from chromosome arms by unordered alpha satellite monomers and other repetitive elements. Complexities in assembling such large repetitive regions have limited detailed studies of centromeric chromatin organization. However, a genomic map of the human X centromere has provided new opportunities to explore genomic architecture of a complex locus. We used ChIP to examine the distribution of modified histones within centromere regions of multiple X chromosomes. Methylation of H3 at lysine 4 coincided with DXZ1 higher order alpha satellite, the site of CENP-A localization. Heterochromatic histone modifications were distributed across the 400-500 kb pericentromeric regions. The large arrays of alpha satellite and gamma satellite DNA were enriched for both euchromatic and heterochromatic modifications, implying that some pericentromeric repeats have multiple chromatin characteristics. Partial truncation of the X centromere resulted in reduction in the size of the CENP-A/Cenp-A domain and increased heterochromatic modifications in the flanking pericentromere. Although the deletion removed approximately 1/3 of centromeric DNA, the ratio of CENP-A to alpha satellite array size was maintained in the same proportion, suggesting that a limited, but defined linear region of the centromeric DNA is necessary for kinetochore assembly. Our results indicate that the human X centromere contains multiple types of chromatin, is organized similarly to smaller eukaryotic centromeres, and responds to structural changes by expanding or contracting domains.

Cysteine proteinase-1 and cut protein isoform control dendritic innervation of two distinct sensory fields by a single neuron.

Dendrites often exhibit structural changes in response to local inputs. Although mechanisms that pattern and maintain dendritic arbors are becoming clearer, processes regulating regrowth, during context-dependent plasticity or after injury, remain poorly understood. We found that a class of Drosophila sensory neurons, through complete pruning and regeneration, can elaborate two distinct dendritic trees, innervating independent sensory fields. An expression screen identified Cysteine proteinase-1 (Cp1) as a critical regulator of this process. Unlike known ecdysone effectors, Cp1-mutant ddaC neurons pruned larval dendrites normally but failed to regrow adult dendrites. Cp1 expression was upregulated/concentrated in the nucleus during metamorphosis, controlling production of a truncated Cut homeodomain transcription factor. This truncated Cut, but not the full-length protein, allowed Cp1-mutant ddaC neurons to regenerate higher-order adult dendrites. These results identify a molecular pathway needed for dendrite regrowth after pruning, which allows the same neuron to innervate distinct sensory fields.

Computational inference of neural information flow networks.

Determining how information flows along anatomical brain pathways is a fundamental requirement for understanding how animals perceive their environments, learn, and behave. Attempts to reveal such neural information flow have been made using linear computational methods, but neural interactions are known to be nonlinear. Here, we demonstrate that a dynamic Bayesian network (DBN) inference algorithm we originally developed to infer nonlinear transcriptional regulatory networks from gene expression data collected with microarrays is also successful at inferring nonlinear neural information flow networks from electrophysiology data collected with microelectrode arrays. The inferred networks we recover from the songbird auditory pathway are correctly restricted to a subset of known anatomical paths, are consistent with timing of the system, and reveal both the importance of reciprocal feedback in auditory processing and greater information flow to higher-order auditory areas when birds hear natural as opposed to synthetic sounds. A linear method applied to the same data incorrectly produces networks with information flow to non-neural tissue and over paths known not to exist. To our knowledge, this study represents the first biologically validated demonstration of an algorithm to successfully infer neural information flow networks.

Specialized motor-driven dusp1 expression in the song systems of multiple lineages of vocal learning birds.

Mechanisms for the evolution of convergent behavioral traits are largely unknown. Vocal learning is one such trait that evolved multiple times and is necessary in humans for the acquisition of spoken language. Among birds, vocal learning is evolved in songbirds, parrots, and hummingbirds. Each time similar forebrain song nuclei specialized for vocal learning and production have evolved. This finding led to the hypothesis that the behavioral and neuroanatomical convergences for vocal learning could be associated with molecular convergence. We previously found that the neural activity-induced gene dual specificity phosphatase 1 (dusp1) was up-regulated in non-vocal circuits, specifically in sensory-input neurons of the thalamus and telencephalon; however, dusp1 was not up-regulated in higher order sensory neurons or motor circuits. Here we show that song motor nuclei are an exception to this pattern. The song nuclei of species from all known vocal learning avian lineages showed motor-driven up-regulation of dusp1 expression induced by singing. There was no detectable motor-driven dusp1 expression throughout the rest of the forebrain after non-vocal motor performance. This pattern contrasts with expression of the commonly studied activity-induced gene egr1, which shows motor-driven expression in song nuclei induced by singing, but also motor-driven expression in adjacent brain regions after non-vocal motor behaviors. In the vocal non-learning avian species, we found no detectable vocalizing-driven dusp1 expression in the forebrain. These findings suggest that independent evolutions of neural systems for vocal learning were accompanied by selection for specialized motor-driven expression of the dusp1 gene in those circuits. This specialized expression of dusp1 could potentially lead to differential regulation of dusp1-modulated molecular cascades in vocal learning circuits.

The past, present, and future of human centromere genomics.

The centromere is the chromosomal locus essential for chromosome inheritance and genome stability. Human centromeres are located at repetitive alpha satellite DNA arrays that compose approximately 5% of the genome. Contiguous alpha satellite DNA sequence is absent from the assembled reference genome, limiting current understanding of centromere organization and function. Here, we review the progress in centromere genomics spanning the discovery of the sequence to its molecular characterization and the work done during the Human Genome Project era to elucidate alpha satellite structure and sequence variation. We discuss exciting recent advances in alpha satellite sequence assembly that have provided important insight into the abundance and complex organization of this sequence on human chromosomes. In light of these new findings, we offer perspectives for future studies of human centromere assembly and function.

A natural extension of the conventional finite volume method into polygonal unstructured meshes for CFD application.

A new general cell-centered solution procedure based upon the conventional control or finite volume (CV or FV) approach has been developed for numerical heat transfer and fluid flow which encompasses both structured and unstructured meshes for any kind of mixed polygon cell. Unlike conventional FV methods for structured and block structured meshes and both FV and FE methods for unstructured meshes, the irregular control volume (ICV) method does not require the shape of the element or cell to be predefined because it simply exploits the concept of fluxes across cell faces. That is, the ICV method enables meshes employing mixtures of triangular, quadrilateral, and any other higher order polygonal cells to be exploited using a single solution procedure. The ICV approach otherwise preserves all the desirable features of conventional FV procedures for a structured mesh; in the current implementation, collocation of variables at cell centers is used with a Rhie and Chow interpolation (to suppress pressure oscillation in the flow field) in the context of the SIMPLE pressure correction solution procedure. In fact all other FV structured mesh-based methods may be perceived as a subset of the ICV formulation. The new ICV formulation is benchmarked using two standard computational fluid dynamics (CFD) problems i.e., the moving lid cavity and the natural convection driven cavity. Both cases were solved with a variety of structured and unstructured meshes, the latter exploiting mixed polygonal cell meshes. The polygonal mesh experiments show a higher degree of accuracy for equivalent meshes (in nodal density terms) using triangular or quadrilateral cells; these results may be interpreted in a manner similar to the CUPID scheme used in structured meshes for reducing numerical diffusion for flows with changing direction.