Search

In October 1992, I was awarded funding by the Edinburgh University Enterprise Centre for a ninemonth project entitled ‘The Development of Self-Directed Learning using Computers and Satellite TV’. The aim of the ‘computers’ half of the project was to find ways of getting first-year students to spend more time working by themselves on the quality of their written French.

Vorwort. Abkurzungen. I. Grammatische Termini und Symbole. II. Sprachen und Dialekte. III. Periodika und wissenschaftliche Serien. IV. Weitere Abkurzungen. Bibliographie. I. Autoren. II. Quellen (albanische Texte, Woerterbucher, Wortsammlungen u.a.). I. Zur Einrichtung des etymologischen Glossars. Lemma-Bestand. Aufbau der einzelnen Lemmata. Sonstige Vorbemerkungen. II. Die Vertretung des idg. Phonemsystems im Albanischen. 1. Die silbischen Vokale und Diphthonge. 2. Die Halbvokale. 3. Die Liquiden und Nasale. 4. Der Spirant */s/. 5. Die Laryngale. 6. Die Verschlusslaute. Lemmata. Wortverzeichnis.

Temporal difference learning has been proposed as a model for Pavlovian conditioning, in which an animal learns to predict delivery of reward following presentation of a conditioned stimulus (CS). A key component of this model is a prediction error signal, which, before learning, responds at the time of presentation of reward but, after learning, shifts its response to the time of onset of the CS. In order to test for regions manifesting this signal profile, subjects were scanned using event-related fMRI while undergoing appetitive conditioning with a pleasant taste reward. Regression analyses revealed that responses in ventral striatum and orbitofrontal cortex were significantly correlated with this error signal, suggesting that, during appetitive conditioning, computations described by temporal difference learning are expressed in the human brain.

Neural processing faces three rather different, and perniciously tied, communication problems. First, computation is radically distributed, yet point-to-point interconnections are limited. Second, the bulk of these connections are semantically uniform, lacking differentiation at their targets that could tag particular sorts of information. Third, the brain's structure is relatively fixed, and yet different sorts of input, forms of processing, and rules for determining the output are appropriate under different, and possibly rapidly changing, conditions. Neuromodulators address these problems by their multifarious and broad distribution, by enjoying specialized receptor types in partially specific anatomical arrangements, and by their ability to mold the activity and sensitivity of neurons and the strength and plasticity of their synapses. Here, I offer a computationally focused review of algorithmic and implementational motifs associated with neuromodulators, using decision making in the face of uncertainty as a running example.

Summary Political ideologies summarize dimensions of life that define how a person organizes their public and private behavior, including their attitudes associated with sex, family, education, and personal autonomy [1, 2]. Despite the abstract nature of such sensibilities, fundamental features of political ideology have been found to be deeply connected to basic biological mechanisms [3–7] that may serve to defend against environmental challenges like contamination and physical threat [8–12]. These results invite the provocative claim that neural responses to nonpolitical stimuli (like contaminated food or physical threats) should be highly predictive of abstract political opinions (like attitudes toward gun control and abortion) [13]. We applied a machine-learning method to fMRI data to test the hypotheses that brain responses to emotionally evocative images predict individual scores on a standard political ideology assay. Disgusting images, especially those related to animal-reminder disgust (e.g., mutilated body), generate neural responses that are highly predictive of political orientation even though these neural predictors do not agree with participants’ conscious rating of the stimuli. Images from other affective categories do not support such predictions. Remarkably, brain responses to a single disgusting stimulus were sufficient to make accurate predictions about an individual subject’s political ideology. These results provide strong support for the idea that fundamental neural processing differences that emerge under the challenge of emotionally evocative stimuli may serve to structure political beliefs in ways formerly unappreciated., Highlights • Literature suggests negativity bias might underlie variations in political views • fMRI responses to disgusting images accurately predict political orientation • Self-reports about affective images are not predictive of their political views • Single-stimulus data can reliably classify conservatives from liberals, Ahn et al. show that fMRI responses to disgusting images accurately predict political orientation. The effect is strong enough to elicit good classification of conservatives from liberals from single-stimulus data, suggesting that emotional processes play a much larger role in structuring our abstract political beliefs than we currently believe.

A number of different methods of calculating the coupling efficiency between a semiconductor waveguide and an optical fibre have been studied. The limitations to the simplest methods have been identified. Practical designs of waveguide have been analysed, and a structure with a coupling efficiency of 0.6 dB has been discussed.

Many neural areas, notably, the hippocampus, show structured, dynamical, population behavior such as coordinated oscillations. It has long been observed that such oscillations provide a substrate for representing analog information in the firing phases of neurons relative to the underlying population rhythm. However, it has become increasingly clear that it is essential for neural populations to rep- resent uncertainty about the information they capture, and the substantial recent work on neural codes for uncertainty has omitted any analysis of oscillatory systems. Here, we observe that, since neurons in an oscillatory network need not only fire once in each cycle (or even at all), uncertainty about the analog quantities each neuron represents by its firing phase might naturally be reported through the degree of concentration of the spikes that it fires. We apply this theory to memory in a model of oscillatory associative recall in hippocampal area CA3. Although it is not well treated in the literature, representing and manipulating uncertainty is fundamental to competent memory; our theory enables us to view CA3 as an effective uncertainty-aware, retrieval system.

Recurrent network models of area CA3 in the hippocampus capture faithfully many of the properties of place cells. However, they seem ill suited to explaining the substantial experimental data on place cells in environments with particular visual or geometrical similarities. We show that a model in which the activities of CA3 place cells are determined mainly by modifiable recurrent connections (together with global inhibitory feedback) is capable of reproducing the major classes of behavior that are observed. In visually similar environments, the patterns of place cell activities have the appropriate degree of similarity; after geometric transformations to the environment, the model place fields undergo geometric transformations, and also remapping, induced (or uncovered) directionality and disappearance.

Decision making lies at the very heart of many psychiatric diseases. It is also a central theoretical concern in a wide variety of fields and has undergone detailed, in-depth, analyses. We take as an example Major Depressive Disorder (MDD), applying insights from a Bayesian reinforcement learning framework. We focus on anhedonia and helplessness. Helplessness—a core element in the conceptual- izations of MDD that has lead to major advances in its treatment, pharmacolog- ical and neurobiological understanding—is formalized as a simple prior over the outcome entropy of actions in uncertain environments. Anhedonia, which is an equally fundamental aspect of the disease, is related to the effective reward size. These formulations allow for the design of specific tasks to measure anhedonia and helplessness behaviorally. We show that these behavioral measures capture explicit, questionnaire-based cognitions. We also provide evidence that these tasks may allow classification of subjects into healthy and MDD groups based purely on a behavioural measure and avoiding any verbal reports.

Neuroscience studies of human decision-making abilities commonly involve sub-jects completing a decision-making task while BOLD signals are recorded using fMRI. Hypotheses are tested about which brain regions mediate the effect of past experience, such as rewards, on future actions. One standard approach to this is model-based fMRI data analysis, in which a model is fitted to the behavioral data, i.e., a subject’s choices, and then the neural data are parsed to find brain regions whose BOLD signals are related to the model’s internal signals. However, the internal mechanics of such purely behavioral models are not constrained by the neural data, and therefore might miss or mischaracterize aspects of the brain. To address this limitation, we introduce a new method using recurrent neural network models that are flexible enough to be jointly fitted to the behavioral and neural data. We trained a model so that its internal states were suitably related to neural activity during the task, while at the same time its output predicted the next action a subject would execute. We then used the fitted model to create a novel visualization of the relationship between the activity in brain regions at different times following a reward and the choices the subject subsequently made. Finally, we validated our method using a previously published dataset. We found that the model was able to recover the underlying neural substrates that were discovered by explicit model engineering in the previous work, and also derived new results regarding the temporal pattern of brain activity.

Dopamine is intimately linked with the modes of action of drugs of addiction. However, although its role in the initiation of drug abuse seems relatively uncomplicated, its possible involvement in the development of compulsive drug taking, and indeed vulnerability and relapse, is less clear. We first describe a modern reinforcement learning view of affective control, focusing on the roles for dopamine. We then use this as a framework to sketch various notions of the neuromodulator's possible participation in initiation and compulsion. We end with some pointers towards future theoretical developments.

We develop a theoretical framework that shows how mesencephalic dopamine systems could distribute to their targets a signal that represents information about future expectations. In particular, we show how activity in the cerebral cortex can make predictions about future receipt of reward and how fluctuations in the activity levels of neurons in diffuse dopamine systems above and below baseline levels would represent errors in these predictions that are delivered to cortical and subcortical targets. We present a model for how such errors could be constructed in a real brain that is consistent with physiological results for a subset of dopaminergic neurons located in the ventral tegmental area and surrounding dopaminergic neurons. The theory also makes testable predictions about human choice behavior on a simple decision-making task. Furthermore, we show that, through a simple influence on synaptic plasticity, fluctuations in dopamine release can act to change the predictions in an appropriate manner.

Neurons in the visual cortex require correlated binocular activity during a critical period early in life to develop normal response properties. We present a model for how the disparity selectivity of cortical neurons might arise during development. The model is based on Hebbian mechanisms for plasticity at synapses between geniculocortical neurons and cortical cells. The model is driven by correlated activity in retinal ganglion cells within each eye before birth and additionally between eyes after birth. With no correlations present between the eyes, the cortical model developed only monocular cells. Adding a small amount of correlation between eyes at the beginning of development produced cortical neurons that were entirely binocular and tuned to zero disparity. However, if an initial phase of purely same-eye correlations was followed by a second phase of development that included correlations between eyes, the cortical model became populated with both monocular and binocular cells. Moreover, in the two-phase model, binocular cells tended to be selective for zero disparity, whereas the more monocular cells tended to have nonzero disparity. This relationship between ocular dominance and disparity has been observed in the visual cortex of the cat by other workers. Differences in the relative timing of the two developmental phases could account for the higher proportion of monocular cells found in the visual cortices of other animals.

This paper describes two new methods for modeling the manifolds of digitized images of handwritten digits. The models allow a priori information about the structure of the manifolds to be combined with empirical data. Accurate modeling of the manifolds allows digits to be discriminated using the relative probability densities under the alternative models. One of the methods is grounded in principal components analysis, the other in factor analysis. Both methods are based on locally linear low-dimensional approximations to the underlying data manifold. Links with other methods that model the manifold are discussed.

Classical game theoretic approaches that make strong rationality assumptions have difficulty modeling human behaviour in economic games. We investigate the role of finite levels of iterated reasoning and non-selfish utility functions in a Partially Observable Markov Decision Process model that incorporates game theoretic no- tions of interactivity. Our generative model captures a broad class of characteristic behaviours in a multi-round Investor-Trustee game. We invert the generative pro- cess for a recognition model that is used to classify 200 subjects playing this game against randomly matched opponents.

The elastic net and related algorithms, such as generative topographic mapping, are key methods for discretized dimension-reduction problems. At their heart are priors that specify the expected topological and geometric properties of the maps. However, up to now, only a very small subset of possible priors has been considered. Here we study a much more general family originating from discrete, high-order derivative operators. We show theoretically that the form of the discrete approximation to the derivative used has a crucial influence on the resulting map. Using a new and more powerful iterative elastic net algorithm, we confirm these results empirically, and illustrate how different priors affect the form of simulated ocular dominance columns.

We present a general encoding-decoding framework for interpreting the activity of a population of units. A standard population code interpretation method, the Poisson model, starts from a description as to how a single value of an underlying quantity can generate the activities of each unit in the population. In casting it in the encoding-decoding framework, we find that this model is too restrictive to describe fully the activities of units in population codes in higher processing areas, such as the medial temporal area. Under a more powerful model, the population activity can convey information not only about a single value of some quantity but also about its whole distribution, including its variance, and perhaps even the certainty the system has in the actual presence in the world of the entity generating this quantity. We propose a novel method for forming such probabilistic interpretations of population codes and compare it to the existing method.

It is a commonplace in statistics that uncertainty about parameters drives learning. Indeed one of the most influential models of behavioural learning has uncertainty at its heart. However, many popular theoretical models of learning focus exclusively on error, and ignore uncertainty. Here we review the links between learning and uncertainty from three perspectives: statistical theories such as the Kalman filter, psychological models in which differential attention is paid to stimuli with an effect on the speed of learning associated with those stimuli, and neurobiological data on the influence of the neuromodulators acetylcholine and norepinephrine on learning and inference.

Computational psychiatry is an emerging field that examines phenomena in mental illness using formal techniques from computational neuroscience, mathematical psychology, and machine learning. These techniques can be used in a theory-driven manner to gain insight into neural or cognitive processes and in a data-driven way to identify predictive and explanatory relationships in complex datasets. The approaches complement each other: theory-driven models can be used to infer mechanisms, and the resulting measurements can be used in data-driven approaches for prediction. Recent computational studies have successfully described and measured novel mechanisms in a range of disorders, have framed disorders in new and informative ways, and have identified predictors of treatment response. These methods hold the potential to improve identification of relevant clinical variables and could be superior to classification based on traditional behavioral or neural data alone. However, these promising results have been slow to influence clinical practice or to improve patient outcomes.

Reinforcement learning (RL) has become a dominant computational paradigm for modeling psychological and neural aspects of affectively charged decision-making tasks. RL is normally construed in terms of the interaction between a subject and its environment, with the former emitting actions, and the latter providing stimuli, and appetitive and aversive reinforcement. However, there is recent emphasis on redrawing the boundary between the two, with the organism constructing its own notion of reward, punishment and state, and with internal actions, such as the gating of working memory, being treated on an equal footing with external manipulation of the environment. We review recent work in this area, focusing on cognitive control.

Background. Real-life decisions are often complex because they involve making sequential choices that constrain future options. We have previously shown that to render such multi-step decisions manageable, people ‘prune’ (i.e. selectively disregard) branches of decision trees that contain negative outcomes. We have theorized that sub-optimal pruning contributes to depression by promoting an oversampling of branches that result in unsavoury outcomes, which results in a negatively-biased valuation of the world. However, no study has tested this theory in depressed individuals. Methods. Thirty unmedicated depressed and 31 healthy participants were administered a sequential reinforcement-based decision-making task to determine pruning behaviours, and completed measures of depression and anxiety. Computational, Bayesian and frequentist analyses examined group differences in task performance and relationships between pruning and depressive symptoms. Results. Consistent with prior findings, participants robustly pruned branches of decision trees that began with large losses, regardless of the potential utility of those branches. However, there was no group difference in pruning behaviours. Further, there was no relationship between pruning and levels of depression/anxiety. Conclusions. We found no evidence that sub-optimal pruning is evident in depression. Future research could determine whether maladaptive pruning behaviours are observable in specific sub-groups of depressed patients (e.g. in treatment-resistant individuals), or whether misuse of other heuristics may contribute to depression., Psychological Medicine, 52 (16), ISSN:0033-2917, ISSN:1469-8978

Residual micro-saccades, tremor and fixation errors imply that, on different trials in visual tasks, stimulus arrays are inevitably presented at different positions on the retina. Positional variation is likely to be specially important for tasks involving visual hyperacuity, because of the severe demands that these tasks impose on spatial resolution. In this paper, we show that small positional variations lead to a structural change in the nature of the ideal observer's solution to a hyperacuity-like visual discrimination task such that the optimal discriminator depends quadratically rather than linearly on noisy neural activities. Motivated by recurrent models of early visual processing, we show how a recurrent preprocessor of the noisy activities can produce outputs which, when passed through a linear discriminator, lead to better discrimination even when the positional variations are much larger than the threshold acuity of the task. Since, psychophysically, hyperacuity typically improves greatly over the course of perceptual learning, we discuss our model in the light of results on the speed and nature of learning.

Acknowledgements Introduction 1. The Danaides Vessel: on reading Baudelaire's allegories 2. On certain relations: figures of sexuality in Baudelaire 3. Emma's stories: narrative, repetition and desire in Madame Bovary 4. The autobiography of rhetoric: on reading Rimbaud's Une Saison en enfer 5. False confusions: ficitons of masculine desire in Mallarme's 'L'Apres-midi d'un faune' 6. The trials of authority under Louis Bonaparte Notes Index.