The dorsal horn of the spinal cord is a complex laminar structure integrating exteroceptive signals from the primary afferent fibers into the central nervous system. The laminae of the spinal cord exhibit specialised roles and distinct processes occur across the axes of the dorsal horn. One of the most common in vivo approaches to recording spinal activity is single unit electrophysiology of cells that are believed to be representative of the subjects perception of stimuli. This approach has produced invaluable data but has not progressed in over half a century and fails to account for the specialised processes that occur in each lamina as well as the considerable cellular heterogeneity within and between laminae. In this thesis the use of linear multi-electrode array technology with 16 electrodes spaced 50 μm apart to have a total range of 750 μm that allows for simultaneous recordings across the laminae is developed and validated for the spinal dorsal horns of adult Sprague-Dawley rats. To do this a series of experiments were performed. The placement of the electrode was first optimised by creating a somatotopic map of evoked activity following hindpaw stimulation. A comprehensive series of electrical stimuli designed to induce differential primary afferent fiber activity were then given to establish how well the array could interpret fiber evoked activity. Mechanical and thermal stimulus paradigms were examined to evaluate the spatial distribution of responses across the dorsal horn; after optimisation the responses were then examined in the contexts of acute and chronic pain models. We found that the results of the unsorted multi-spike activity across the array correlated extremely well with predicted responses from single-unit studies in the existing literature. Fiber specific activation restricted along the dorso-ventral axis was detected as was the encoding of mechanical and thermal stimuli that were both innocuous and nocuous in nature. Comparisons between local field potentials and spike activity showed that multi-spike activity represented spinal processing of incoming signals significantly better. Induction of pain models strengthened the argument for the use of the technique by showing that it is capable of being used in both longitudinal and multi-treatment group studies. The approach produced vastly more data than the single-unit technique it builds upon with few drawbacks.

Past research tells us that individuals can infer information about a target’s emotional state and intentions from their facial expressions (Frith & Frith, 2012), a process known as mentalising. More recently, it has been found that this ability extends to inferring the events that caused the facial reaction (e.g. Pillai, Sheppard, & Mitchell, 2012; Pillai et al., 2014), an ability known as retrodictive mindreading. In the current thesis, we enter a new territory where a series of experiments was conducted to investigate whether people (perceivers) can guess a target’s social context by observing their response to emotional stimuli. The core findings were: 1) perceivers were able to discriminate whether the targets were alone or observed by another person, 2) without any knowledge of the social context or what the targets were watching, perceivers judged whether targets were hiding or exaggerating their facial expressions, and their judgments discriminated between conditions in which targets were observed and alone, and 3) perceivers’ eye movements also systematically discriminated between conditions in which targets were observed and alone. Perceivers were thus able to infer – explicitly or implicitly - a target’s social context by observing their emotional response. Therefore, the findings demonstrate that people have the ability to use other people’s minds as a window onto a social context that could not be seen directly.

Internal or Reflective attention can refer to our thoughts/reflections in order to make sense of our external world through our senses and perception. Reflective attention also includes the act of refreshing which is the act of thinking back and shifting internal attention towards previously activated mental representations. Previous research (M.R. Johnson et al., 2013) has shown that refreshing mirrors a striking similarity to that of inhibition of return (IOR) effect which inhibits visual attention to return to a previously cued location (Posner & Cohen, 1984; Posner, Rafal, Choate, & Vaughan, 1985). This IOR-like mechanism helps facilitate our thoughts (similarly to perception) by encouraging internal attention to move towards new information and avoid constant fixation on a single thought (M.R. Johnson et al., 2013) which was coined as reflective IOR (rIOR). The objective of the thesis investigates variables such as time duration and language during the production of rIOR mechanism. A total of seven experiments were conducted. The first set of experiments (Experiments 1 to 3) aimed to examine the time course of refreshing while the second set and (Experiments 4 to 7) examined the effect of language on reflective attention. In each experiment, participants were shown two stimuli, either in the form of pictures or English/Malay words. They were instructed to refresh by keeping one item (i.e., mental representation) active while ignoring the other. Results showed an attentional shift or bias towards the unrefreshed mental representation, more so in the experiments which used word stimuli rather than picture stimuli. The novelty of the current thesis is that early language processing (i.e., English and Malay words) in bilingual speakers was taken into account while investigating the reflective attention. This pattern was consistent whether the words were presented in English or Malay which are consistent with M. R. Johnson and colleagues’ finding that IOR mechanism shifts internal attention to new information However, if participants were presented with English stimuli, refreshed the English word but were then probed in the equivalent Malay word, a stronger priming effect emerged instead. The behavioural pattern implicated that asymmetrical cost during language switching could be reduced as a result of refreshing. The data also showed that while refreshing may cause a temporary inaccessibility to recently activated items, refreshed words were more memorable in a later recognition task. This suggested the role of refreshing plays an important role encoding and storing mental representations in later long term retrieval. Mental representations that were ignored or were not give n attention tended to fade away more quickly. The novelty of the thesis is that language processing was explored as a component to this mechanism by manipulating languages of the refreshed words presented. Participants were more likely to make false alarms when they were presented with an English equivalent word in the recognition task, when the original word had in fact been presented (i.e., previously refreshed) in Malay. Language models such as the Revised Hierarchical Model (RHM) ( Kroll and Stewart, 1990, 1994) were applied in examining refreshing in stronger or weaker languages that gave rise to poor memory performance. According to the RHM’s logic, words activated in non-dominant language would subsequently activate words in the dominant language in order to access the meaning of the word. In this language processing route, it is possible that refreshing a word in the weaker language would subsequently activate the similar word in the stronger language which is reflected as a false memory incident.

Recent developments in the embodied approach to understanding the generation of adaptive behaviour, suggests that the design of adaptive neural circuits for rhythmic motor patterns should not be done in isolation from an appreciation, and indeed exploitation, of neural-body-environment interactions. Utilising spontaneous mutual entrainment between neural systems and physical bodies provides a useful passage to the regions of phase space which are naturally structured by the neuralbody- environmental interactions. A growing body of work has provided evidence that chaotic dynamics can be useful in allowing embodied systems to spontaneously explore potentially useful motor patterns. However, up until now there has been no general integrated neural system that allows goal-directed, online, realtime exploration and capture of motor patterns without recourse to external monitoring, evaluation or training methods. For the first time, we introduce such a system in the form of a fully dynamic neural system, exploiting intrinsic chaotic dynamics, for the exploration and learning of the possible locomotion patterns of an articulated robot of an arbitrary morphology in an unknown environment. The controller is modelled as a network of neural oscillators which are coupled only through physical embodiment, and goal directed exploration of coordinated motor patterns is achieved by a chaotic search using adaptive bifurcation. The phase space of the indirectly coupled neural-body-environment system contains multiple transient or permanent self-organised dynamics each of which is a candidate for a locomotion behaviour. The adaptive bifurcation enables the system orbit to wander through various phase-coordinated states using its intrinsic chaotic dynamics as a driving force and stabilises the system on to one of the states matching the given goal criteria. In order to improve the sustainability of useful transient patterns, sensory homeostasis has been introduced which results in an increased diversity of motor outputs, thus achieving multi-scale exploration. A rhythmic pattern discovered by this process is memorised and sustained by changing the wiring between initially disconnected oscillators using an adaptive synchronisation method. The dynamical nature of the weak coupling through physical embodiment allows this adaptive weight learning to be easily integrated, thus forming a continuous exploration-learning system. Our result shows that the novel neuro-robotic system is able to create and learn a number of emergent locomotion behaviours for a wide range of body configurations and physical environment, and can re-adapt after sustaining damage. The implications and analyses of these results for investigating the generality and limitations of the proposed system are discussed.

The GABAA α4-subunit is found co-assembled with δ subunits in extrasynaptic GABAA receptors (α4-GABAARs). Within the striatum α4-GABAARs are most highly expressed in the Nucleus Accumbens (NAc) where they mediate tonic inhibition thought to control the excitability of accumbal medium spiny neurons (MSNs). Experiments presented in this thesis use genetic techniques in mice to investigate the role of α4-GABAARs in modulating binge-like ethanol consumption and the potentiation of locomotor behaviours by cocaine. We have generated several transgenic mouse lines in which the Gabrα4 gene, encoding the α4 subunit, has been deleted either constitutively or within specific neural populations expressing D1 or D2 type dopamine receptors via cre/loxp recombination. Using quantitative rt-PCR and in-situ-hybridisation methods to compare Gabrα4 mRNA levels in brain sections from each genotype we confirmed that the α4 subunit was deleted either globally or in the expected cell type within conditional knockouts. We also generated an Adeno Associated Virus (AAV) carrying Cre-recombinase to knockdown α4 locally by infusing it into in specific brain regions of ‘floxed'-α4 mice. Deletion of the α4 subunit in mice significantly reduced alcohol consumption in a pre-clinical model of binge-drinking, known as drinking in the dark (DID). Moreover, targeted deletion of Gabrα4 in the NAc was sufficient to mediate this effect. We did not observe any effects on alcohol consumption in mice where α4 was deleted conditionally in D1 or D2 type neurons. This data indicates that α4- GABAARs in the NAc are an important mediator of alcohol consumption. Deletion of GABAAR α4-subunits from dopamine D1-expressing neurons facilitated cocaine's ability to potentiate locomotor activity and operant responding for natural rewards. Deletion of GABAAR α4-subunits from dopamine D2-expressing neurons had no such effects. Deletion of GABAAR α4-subunits from dopamine D1- expressing neurons also accelerated the acquisition of behavioural sensitisation to cocaine. This effect was associated with increased cFos expression in the NAc core following acute cocaine, whilst in cocaine-sensitised mice it was associated with increased cFos in both the NAc Core and Shell. A similar altered pattern of cFos expression was observed in mice with a global knockout of α4 subunits however they showed no behavioural effects. This may imply that a balance of α4-GABAARmediated inhibition in D1 and D2 neurons is required for normal behavioural sensitisation to cocaine. The data presented within this thesis indicate that α4- GABAAR-mediated inhibition of D1- and D2-expressing neurons plays an important physiological role in controlling behavioural responses to cocaine.

The interaction of entities on a network structure is of significant importance to many disciplines. Network structures can have both physical (e.g. power grids, computer networks, the World Wide Web, networks of neurones) and non-physical (e.g. social networks of friends, links between communities, the movement of livestock) realisations that are all amenable to study. In this thesis work on dynamical processes and the networks on which they occur is presented from a viewpoint of both mathematical epidemiology and computational/theoretical neuroscience, with additional consideration of the intersection between the two. I begin with a paper illustrating how different models of disease transmission are derivable from others and provide a framework for the development of approximate ODEs based on their derivation from exact Kolmogorov equations. This work is followed with two papers that use two such approximate models and consider how they perform when the interplay between both disease and network dynamics is taken into account. Whilst the work in these papers focusses on the modelling of the temporal evolution of the disease and network dynamics, papers four and five consider the recent viewpoint within neuroscience that the brain operates within a critical regime. Making use of models analogous to meanfield models in epidemiology I analyse the behaviour of the system when it is in a balanced state, characterised by the system operating at or near its critical bifurcation, and how this is relevant to the brain itself. Whilst models used within the two areas are analogous, the behavioural aspects of interest within them are quite different. I conclude with a discussion of these differences, the overlaps between both fields and suggest where future work in each area may benefit from incorporating methods and ideas of the other.

Much behaviour is controlled by neural circuits known as central pattern generators (CPGs). The aim of the work presented in this thesis was to uncover general mechanisms that modify the behavioural output of CPGs in ways that maximise adaptive fitness. To achieve this aim it was necessary to monitor populations of neurons associated with a CPG that responds to changes in sensory reward. I used multi-electrode arrays (MEAs) to monitor neuronal populations in semi-intact preparations of the snail Lymnaea stagnalis. Spike patterns associated with cycles of the feeding CPG were readily recorded in the buccal, cerebral and pedal ganglia. A sensory food stimulus accelerated the CPG and this acceleration was shown to depend on dopamine. Single-trial conditioning on the MEA allowed fictive feeding to be induced by a previously neutral taste stimulus. In addition to the activity of the feeding CPG the MEA also revealed a second neuronal population that had not previously been characterized. This population fires continuously in-between the cycles of the feeding CPG but becomes quiescent for a variable period following each cycle. The duration of this quiescent period often predicted the timing of the next activation of the CPG. Stimulation of a nerve associated with food reward failed to activate the CPG during the quiescent period, indicating that it reflects a ‘network refractory period' (NRP) of the kind previously observed in locomotor CPGs. The sucrose and dopamine stimuli both significantly shortened the NRP. These results show that the MEA recording method can identify distinct populations of neurons associated with adaptive feeding behaviour, and suggest a general mechanism that allows a CPG to adapt its behavioural output to maximise reward

Why does it feel like something to be awake? I.e. how is consciousness generated by the body, the brain in particular? Seeking to map phenomenological properties of any first person experience to neural activity patterns, theories of consciousness suggest a correlation between a specific type of neural dynamical complexity and the level of consciousness: When awake and aware, all brain regions are to a certain extent connected and there is diversity in the interactions. In support of this, Casali et al. (2013) have used EEG and transcranial magnetic stimulation to show extensively that brain response activity to direct perturbation is the more diverse across regions and time, the higher the level of consciousness. The spatio-temporal diversity of the response signal is quantified by a single index, the perturbational complexity index (PCI), using a Lempel-Ziv compression algorithm. Motivated by this result, and given that spontaneous neural signals are easier to obtain than response signals to perturbation, this thesis proposes measures - based on Lempel-Ziv compression and entropy - to quantify spontaneous neural signal diversity across channels and observations. Our measures' sensitivity and specificity to conscious level is demonstrated by re-analysing resting state scalp EEG during propofol-induced anaesthesia and depth electrode recordings during sleep stages, resulting in consistently higher scores for subjects that are awake than being in propofol-induced anaesthesia or non-rapid eye movement sleep. In addition we demonstrate that our measures score higher for states induced by psychedelic substances by re-analysing resting state magnetoencephalography (MEG) data. We further explore in computer simulation how our measures and PCI behave as a function of connectivity of coupled oscillators, informing models of brain mechanisms associated with the loss of consciousness. While our measures may be weaker than PCI in terms of specificity and sensitivity to conscious level, they are quick and easy to compute and applicable to readily available resting state data. This thesis provides strong evidence that cortical signal diversity is a hallmark of consciousness, as predicted by integrated information and complexity theories of consciousness.

The late 20th century brought with it substantial evidence showing that many vertebrate species are capable of exhibiting brain and behavioural lateralisation (i.e. functional and/or structural specialisations of the left and right sides of the brain/behaviour), undermining the traditional view that this attribute is uniquely human. Motor lateralisation, a specific form of behavioural lateralisation, in which an organism displays a directional preference whilst making movements, has previously been identified in many vertebrates but has received little attention in invertebrate species, particularly insects. Considering this, I investigated motor lateralisation in the desert locust (Schistocerca gregaria) and the red wood ant (Formica rufa) using a combination of a gap-crossing paradigm and a ‘Y'-maze choice experiment. Using these paradigms, I show that the relatively small nervous systems of insects are capable of producing lateralised motor behaviour, providing evidence that being strongly lateralised can be advantageous and, more generally, adding support to the hypothesis that social organisms are more likely to align their lateralisation with others in their group. The gap-crossing paradigm demonstrates that both desert locusts and red wood ants display a forelimb preference, the first direct evidence that an arthropod possesses a preference in the use of otherwise symmetrical limbs. Studying locusts in this paradigm also produced the first evidence that strong lateralisation confers an advantage to the individual because more strongly biased locusts display fewer reaching errors. In addition, wood ants display inter-colony variation in their preference, the first demonstration of a previously undescribed form of lateralisaton, colony-level lateralisation. By testing wood ants over both the gap-crossing and ‘Y'-maze paradigms I show that within the same insect species, lateralisation can be expressed in different forms of motor behaviour.

Systems neuroscience has recently unveiled numerous fundamental features of the macroscopic architecture of the human brain, the connectome, and we are beginning to understand how characteristics of brain dynamics emerge from the underlying anatomical connectivity. The current work utilises complex network analysis on a high-resolution structural connectivity of the human cortex to identify generic organisation principles, such as centralised, modular and hierarchical properties, as well as specific areas that are pivotal in shaping cortical dynamics and function. After confirming its small-world and modular architecture, we characterise the cortex' multilevel modular hierarchy, which appears to be reasonably centralised towards the brain's strong global structural core. The potential functional importance of the core and hub regions is assessed by various complex network metrics, such as integration measures, network vulnerability and motif spectrum analysis. Dynamics facilitated by the large-scale cortical topology is explored by simulating coupled oscillators on the anatomical connectivity. The results indicate that cortical connectivity appears to favour high dynamical complexity over high synchronizability. Taking the ability to entrain other brain regions as a proxy for the threat posed by a potential epileptic focus in a given region, we also show that epileptic foci in topologically more central areas should pose a higher epileptic threat than foci in more peripheral areas. To assess the influence of macroscopic brain anatomy in shaping global resting state dynamics on slower time scales, we compare empirically obtained functional connectivity data with data from simulating dynamics on the structural connectivity. Despite considerable micro-scale variability between the two functional connectivities, our simulations are able to approximate the profile of the empirical functional connectivity. Our results outline the combined characteristics a hierarchically modular and reasonably centralised macroscopic architecture of the human cerebral cortex, which, through these topological attributes, appears to facilitate highly complex dynamics and fundamentally shape brain function.