Implementing statistical learning methods through Bayesian networks (Part 2): bayesian evaluations for results of black toner analyses in forensic document examination

This paper presents and discusses the use of Bayesian procedures - introduced through the use of Bayesian networks in Part I of this series of papers - for 'learning' probabilities from data. The discussion will relate to a set of real data on characteristics of black toners commonly used in printing and copying devices. Particular attention is drawn to the incorporation of the proposed procedures as an integral part in probabilistic inference schemes (notably in the form of Bayesian networks) that are intended to address uncertainties related to particular propositions of interest (e.g., whether or not a sample originates from a particular source). The conceptual tenets of the proposed methodologies are presented along with aspects of their practical implementation using currently available Bayesian network software.

Statistical learning theory for geospatial data. Case study: Aral sea

In recent years there has been an explosive growth in the development of adaptive and data driven methods. One of the efficient and data-driven approaches is based on statistical learning theory (Vapnik 1998). The theory is based on Structural Risk Minimisation (SRM) principle and has a solid statistical background. When applying SRM we are trying not only to reduce training error ? to fit the available data with a model, but also to reduce the complexity of the model and to reduce generalisation error. Many nonlinear learning procedures recently developed in neural networks and statistics can be understood and interpreted in terms of the structural risk minimisation inductive principle. A recent methodology based on SRM is called Support Vector Machines (SVM). At present SLT is still under intensive development and SVM find new areas of application (www.kernel-machines.org). SVM develop robust and non linear data models with excellent generalisation abilities that is very important both for monitoring and forecasting. SVM are extremely good when input space is high dimensional and training data set i not big enough to develop corresponding nonlinear model. Moreover, SVM use only support vectors to derive decision boundaries. It opens a way to sampling optimization, estimation of noise in data, quantification of data redundancy etc. Presentation of SVM for spatially distributed data is given in (Kanevski and Maignan 2004).

Implementing statistical learning methods through Bayesian networks. Part 1: a guide to Bayesian parameter estimation using forensic science data

As a thorough aggregation of probability and graph theory, Bayesian networks currently enjoy widespread interest as a means for studying factors that affect the coherent evaluation of scientific evidence in forensic science. Paper I of this series of papers intends to contribute to the discussion of Bayesian networks as a framework that is helpful for both illustrating and implementing statistical procedures that are commonly employed for the study of uncertainties (e.g. the estimation of unknown quantities). While the respective statistical procedures are widely described in literature, the primary aim of this paper is to offer an essentially non-technical introduction on how interested readers may use these analytical approaches - with the help of Bayesian networks - for processing their own forensic science data. Attention is mainly drawn to the structure and underlying rationale of a series of basic and context-independent network fragments that users may incorporate as building blocs while constructing larger inference models. As an example of how this may be done, the proposed concepts will be used in a second paper (Part II) for specifying graphical probability networks whose purpose is to assist forensic scientists in the evaluation of scientific evidence encountered in the context of forensic document examination (i.e. results of the analysis of black toners present on printed or copied documents).

Contextual factors multiplex to control multisensory processes.

This study analyzed high-density event-related potentials (ERPs) within an electrical neuroimaging framework to provide insights regarding the interaction between multisensory processes and stimulus probabilities. Specifically, we identified the spatiotemporal brain mechanisms by which the proportion of temporally congruent and task-irrelevant auditory information influences stimulus processing during a visual duration discrimination task. The spatial position (top/bottom) of the visual stimulus was indicative of how frequently the visual and auditory stimuli would be congruent in their duration (i.e., context of congruence). Stronger influences of irrelevant sound were observed when contexts associated with a high proportion of auditory-visual congruence repeated and also when contexts associated with a low proportion of congruence switched. Context of congruence and context transition resulted in weaker brain responses at 228 to 257 ms poststimulus to conditions giving rise to larger behavioral cross-modal interactions. Importantly, a control oddball task revealed that both congruent and incongruent audiovisual stimuli triggered equivalent non-linear multisensory interactions when congruence was not a relevant dimension. Collectively, these results are well explained by statistical learning, which links a particular context (here: a spatial location) with a certain level of top-down attentional control that further modulates cross-modal interactions based on whether a particular context repeated or changed. The current findings shed new light on the importance of context-based control over multisensory processing, whose influences multiplex across finer and broader time scales.

Analysis, modelling and classification of geospatial data using machine learning

The research considers the problem of spatial data classification using machine learning algorithms: probabilistic neural networks (PNN) and support vector machines (SVM). As a benchmark model simple k-nearest neighbor algorithm is considered. PNN is a neural network reformulation of well known nonparametric principles of probability density modeling using kernel density estimator and Bayesian optimal or maximum a posteriori decision rules. PNN is well suited to problems where not only predictions but also quantification of accuracy and integration of prior information are necessary. An important property of PNN is that they can be easily used in decision support systems dealing with problems of automatic classification. Support vector machine is an implementation of the principles of statistical learning theory for the classification tasks. Recently they were successfully applied for different environmental topics: classification of soil types and hydro-geological units, optimization of monitoring networks, susceptibility mapping of natural hazards. In the present paper both simulated and real data case studies (low and high dimensional) are considered. The main attention is paid to the detection and learning of spatial patterns by the algorithms applied.

Segmentation and Clustering of Textual Sequences: a Typological Approach

The long term goal of this research is to develop a program able to produce an automatic segmentation and categorization of textual sequences into discourse types. In this preliminary contribution, we present the construction of an algorithm which takes a segmented text as input and attempts to produce a categorization of sequences, such as narrative, argumentative, descriptive and so on. Also, this work aims at investigating a possible convergence between the typological approach developed in particular in the field of text and discourse analysis in French by Adam (2008) and Bronckart (1997) and unsupervised statistical learning.

Mapping of Environmental Data Using Kernel-Based Methods

Recently, kernel-based Machine Learning methods have gained great popularity in many data analysis and data mining fields: pattern recognition, biocomputing, speech and vision, engineering, remote sensing etc. The paper describes the use of kernel methods to approach the processing of large datasets from environmental monitoring networks. Several typical problems of the environmental sciences and their solutions provided by kernel-based methods are considered: classification of categorical data (soil type classification), mapping of environmental and pollution continuous information (pollution of soil by radionuclides), mapping with auxiliary information (climatic data from Aral Sea region). The promising developments, such as automatic emergency hot spot detection and monitoring network optimization are discussed as well.

Automatic classification of MR scans in Alzheimer's disease.

To be diagnostically useful, structural MRI must reliably distinguish Alzheimer's disease (AD) from normal aging in individual scans. Recent advances in statistical learning theory have led to the application of support vector machines to MRI for detection of a variety of disease states. The aims of this study were to assess how successfully support vector machines assigned individual diagnoses and to determine whether data-sets combined from multiple scanners and different centres could be used to obtain effective classification of scans. We used linear support vector machines to classify the grey matter segment of T1-weighted MR scans from pathologically proven AD patients and cognitively normal elderly individuals obtained from two centres with different scanning equipment. Because the clinical diagnosis of mild AD is difficult we also tested the ability of support vector machines to differentiate control scans from patients without post-mortem confirmation. Finally we sought to use these methods to differentiate scans between patients suffering from AD from those with frontotemporal lobar degeneration. Up to 96% of pathologically verified AD patients were correctly classified using whole brain images. Data from different centres were successfully combined achieving comparable results from the separate analyses. Importantly, data from one centre could be used to train a support vector machine to accurately differentiate AD and normal ageing scans obtained from another centre with different subjects and different scanner equipment. Patients with mild, clinically probable AD and age/sex matched controls were correctly separated in 89% of cases which is compatible with published diagnosis rates in the best clinical centres. This method correctly assigned 89% of patients with post-mortem confirmed diagnosis of either AD or frontotemporal lobar degeneration to their respective group. Our study leads to three conclusions: Firstly, support vector machines successfully separate patients with AD from healthy aging subjects. Secondly, they perform well in the differential diagnosis of two different forms of dementia. Thirdly, the method is robust and can be generalized across different centres. This suggests an important role for computer based diagnostic image analysis for clinical practice.

Kernel-based Mapping of Meteorological Fields in Complex Orography

Due to the advances in sensor networks and remote sensing technologies, the acquisition and storage rates of meteorological and climatological data increases every day and ask for novel and efficient processing algorithms. A fundamental problem of data analysis and modeling is the spatial prediction of meteorological variables in complex orography, which serves among others to extended climatological analyses, for the assimilation of data into numerical weather prediction models, for preparing inputs to hydrological models and for real time monitoring and short-term forecasting of weather.In this thesis, a new framework for spatial estimation is proposed by taking advantage of a class of algorithms emerging from the statistical learning theory. Nonparametric kernel-based methods for nonlinear data classification, regression and target detection, known as support vector machines (SVM), are adapted for mapping of meteorological variables in complex orography.With the advent of high resolution digital elevation models, the field of spatial prediction met new horizons. In fact, by exploiting image processing tools along with physical heuristics, an incredible number of terrain features which account for the topographic conditions at multiple spatial scales can be extracted. Such features are highly relevant for the mapping of meteorological variables because they control a considerable part of the spatial variability of meteorological fields in the complex Alpine orography. For instance, patterns of orographic rainfall, wind speed and cold air pools are known to be correlated with particular terrain forms, e.g. convex/concave surfaces and upwind sides of mountain slopes.Kernel-based methods are employed to learn the nonlinear statistical dependence which links the multidimensional space of geographical and topographic explanatory variables to the variable of interest, that is the wind speed as measured at the weather stations or the occurrence of orographic rainfall patterns as extracted from sequences of radar images. Compared to low dimensional models integrating only the geographical coordinates, the proposed framework opens a way to regionalize meteorological variables which are multidimensional in nature and rarely show spatial auto-correlation in the original space making the use of classical geostatistics tangled.The challenges which are explored during the thesis are manifolds. First, the complexity of models is optimized to impose appropriate smoothness properties and reduce the impact of noisy measurements. Secondly, a multiple kernel extension of SVM is considered to select the multiscale features which explain most of the spatial variability of wind speed. Then, SVM target detection methods are implemented to describe the orographic conditions which cause persistent and stationary rainfall patterns. Finally, the optimal splitting of the data is studied to estimate realistic performances and confidence intervals characterizing the uncertainty of predictions.The resulting maps of average wind speeds find applications within renewable resources assessment and opens a route to decrease the temporal scale of analysis to meet hydrological requirements. Furthermore, the maps depicting the susceptibility to orographic rainfall enhancement can be used to improve current radar-based quantitative precipitation estimation and forecasting systems and to generate stochastic ensembles of precipitation fields conditioned upon the orography.

Robust Nonlinear Mapping of Soil Contamination Using Support Regression

Spatial data analysis mapping and visualization is of great importance in various fields: environment, pollution, natural hazards and risks, epidemiology, spatial econometrics, etc. A basic task of spatial mapping is to make predictions based on some empirical data (measurements). A number of state-of-the-art methods can be used for the task: deterministic interpolations, methods of geostatistics: the family of kriging estimators (Deutsch and Journel, 1997), machine learning algorithms such as artificial neural networks (ANN) of different architectures, hybrid ANN-geostatistics models (Kanevski and Maignan, 2004; Kanevski et al., 1996), etc. All the methods mentioned above can be used for solving the problem of spatial data mapping. Environmental empirical data are always contaminated/corrupted by noise, and often with noise of unknown nature. That's one of the reasons why deterministic models can be inconsistent, since they treat the measurements as values of some unknown function that should be interpolated. Kriging estimators treat the measurements as the realization of some spatial randomn process. To obtain the estimation with kriging one has to model the spatial structure of the data: spatial correlation function or (semi-)variogram. This task can be complicated if there is not sufficient number of measurements and variogram is sensitive to outliers and extremes. ANN is a powerful tool, but it also suffers from the number of reasons. of a special type ? multiplayer perceptrons ? are often used as a detrending tool in hybrid (ANN+geostatistics) models (Kanevski and Maignank, 2004). Therefore, development and adaptation of the method that would be nonlinear and robust to noise in measurements, would deal with the small empirical datasets and which has solid mathematical background is of great importance. The present paper deals with such model, based on Statistical Learning Theory (SLT) - Support Vector Regression. SLT is a general mathematical framework devoted to the problem of estimation of the dependencies from empirical data (Hastie et al, 2004; Vapnik, 1998). SLT models for classification - Support Vector Machines - have shown good results on different machine learning tasks. The results of SVM classification of spatial data are also promising (Kanevski et al, 2002). The properties of SVM for regression - Support Vector Regression (SVR) are less studied. First results of the application of SVR for spatial mapping of physical quantities were obtained by the authorsin for mapping of medium porosity (Kanevski et al, 1999), and for mapping of radioactively contaminated territories (Kanevski and Canu, 2000). The present paper is devoted to further understanding of the properties of SVR model for spatial data analysis and mapping. Detailed description of the SVR theory can be found in (Cristianini and Shawe-Taylor, 2000; Smola, 1996) and basic equations for the nonlinear modeling are given in section 2. Section 3 discusses the application of SVR for spatial data mapping on the real case study - soil pollution by Cs137 radionuclide. Section 4 discusses the properties of the modelapplied to noised data or data with outliers.

Electrical neuroimaging of memory discrimination based on single-trial multisensory learning.

Multisensory experiences influence subsequent memory performance and brain responses. Studies have thus far concentrated on semantically congruent pairings, leaving unresolved the influence of stimulus pairing and memory sub-types. Here, we paired images with unique, meaningless sounds during a continuous recognition task to determine if purely episodic, single-trial multisensory experiences can incidentally impact subsequent visual object discrimination. Psychophysics and electrical neuroimaging analyses of visual evoked potentials (VEPs) compared responses to repeated images either paired or not with a meaningless sound during initial encounters. Recognition accuracy was significantly impaired for images initially presented as multisensory pairs and could not be explained in terms of differential attention or transfer of effects from encoding to retrieval. VEP modulations occurred at 100-130ms and 270-310ms and stemmed from topographic differences indicative of network configuration changes within the brain. Distributed source estimations localized the earlier effect to regions of the right posterior temporal gyrus (STG) and the later effect to regions of the middle temporal gyrus (MTG). Responses in these regions were stronger for images previously encountered as multisensory pairs. Only the later effect correlated with performance such that greater MTG activity in response to repeated visual stimuli was linked with greater performance decrements. The present findings suggest that brain networks involved in this discrimination may critically depend on whether multisensory events facilitate or impair later visual memory performance. More generally, the data support models whereby effects of multisensory interactions persist to incidentally affect subsequent behavior as well as visual processing during its initial stages.

The behavioral relevance and top-down susceptibility of early, low level multisensory interactions in humans

Abstract (English)General backgroundMultisensory stimuli are easier to recognize, can improve learning and a processed faster compared to unisensory ones. As such, the ability an organism has to extract and synthesize relevant sensory inputs across multiple sensory modalities shapes his perception of and interaction with the environment. A major question in the scientific field is how the brain extracts and fuses relevant information to create a unified perceptual representation (but also how it segregates unrelated information). This fusion between the senses has been termed "multisensory integration", a notion that derives from seminal animal single-cell studies performed in the superior colliculus, a subcortical structure shown to create a multisensory output differing from the sum of its unisensory inputs. At the cortical level, integration of multisensory information is traditionally deferred to higher classical associative cortical regions within the frontal, temporal and parietal lobes, after extensive processing within the sensory-specific and segregated pathways. However, many anatomical, electrophysiological and neuroimaging findings now speak for multisensory convergence and interactions as a distributed process beginning much earlier than previously appreciated and within the initial stages of sensory processing.The work presented in this thesis is aimed at studying the neural basis and mechanisms of how the human brain combines sensory information between the senses of hearing and touch. Early latency non-linear auditory-somatosensory neural response interactions have been repeatedly observed in humans and non-human primates. Whether these early, low-level interactions are directly influencing behavioral outcomes remains an open question as they have been observed under diverse experimental circumstances such as anesthesia, passive stimulation, as well as speeded reaction time tasks. Under laboratory settings, it has been demonstrated that simple reaction times to auditory-somatosensory stimuli are facilitated over their unisensory counterparts both when delivered to the same spatial location or not, suggesting that audi- tory-somatosensory integration must occur in cerebral regions with large-scale spatial representations. However experiments that required the spatial processing of the stimuli have observed effects limited to spatially aligned conditions or varying depending on which body part was stimulated. Whether those divergences stem from task requirements and/or the need for spatial processing has not been firmly established.Hypotheses and experimental resultsIn a first study, we hypothesized that auditory-somatosensory early non-linear multisensory neural response interactions are relevant to behavior. Performing a median split according to reaction time of a subset of behavioral and electroencephalographic data, we found that the earliest non-linear multisensory interactions measured within the EEG signal (i.e. between 40-83ms post-stimulus onset) were specific to fast reaction times indicating a direct correlation of early neural response interactions and behavior.In a second study, we hypothesized that the relevance of spatial information for task performance has an impact on behavioral measures of auditory-somatosensory integration. Across two psychophysical experiments we show that facilitated detection occurs even when attending to spatial information, with no modulation according to spatial alignment of the stimuli. On the other hand, discrimination performance with probes, quantified using sensitivity (d'), is impaired following multisensory trials in general and significantly more so following misaligned multisensory trials.In a third study, we hypothesized that behavioral improvements might vary depending which body part is stimulated. Preliminary results suggest a possible dissociation between behavioral improvements andERPs. RTs to multisensory stimuli were modulated by space only in the case when somatosensory stimuli were delivered to the neck whereas multisensory ERPs were modulated by spatial alignment for both types of somatosensory stimuli.ConclusionThis thesis provides insight into the functional role played by early, low-level multisensory interac-tions. Combining psychophysics and electrical neuroimaging techniques we demonstrate the behavioral re-levance of early and low-level interactions in the normal human system. Moreover, we show that these early interactions are hermetic to top-down influences on spatial processing suggesting their occurrence within cerebral regions having access to large-scale spatial representations. We finally highlight specific interactions between auditory space and somatosensory stimulation on different body parts. Gaining an in-depth understanding of how multisensory integration normally operates is of central importance as it will ultimately permit us to consider how the impaired brain could benefit from rehabilitation with multisensory stimula-Abstract (French)Background théoriqueDes stimuli multisensoriels sont plus faciles à reconnaître, peuvent améliorer l'apprentissage et sont traités plus rapidement comparé à des stimuli unisensoriels. Ainsi, la capacité qu'un organisme possède à extraire et à synthétiser avec ses différentes modalités sensorielles des inputs sensoriels pertinents, façonne sa perception et son interaction avec l'environnement. Une question majeure dans le domaine scientifique est comment le cerveau parvient à extraire et à fusionner des stimuli pour créer une représentation percep- tuelle cohérente (mais aussi comment il isole les stimuli sans rapport). Cette fusion entre les sens est appelée "intégration multisensorielle", une notion qui provient de travaux effectués dans le colliculus supérieur chez l'animal, une structure sous-corticale possédant des neurones produisant une sortie multisensorielle différant de la somme des entrées unisensorielles. Traditionnellement, l'intégration d'informations multisen- sorielles au niveau cortical est considérée comme se produisant tardivement dans les aires associatives supérieures dans les lobes frontaux, temporaux et pariétaux, suite à un traitement extensif au sein de régions unisensorielles primaires. Cependant, plusieurs découvertes anatomiques, électrophysiologiques et de neuroimageries remettent en question ce postulat, suggérant l'existence d'une convergence et d'interactions multisensorielles précoces.Les travaux présentés dans cette thèse sont destinés à mieux comprendre les bases neuronales et les mécanismes impliqués dans la combinaison d'informations sensorielles entre les sens de l'audition et du toucher chez l'homme. Des interactions neuronales non-linéaires précoces audio-somatosensorielles ont été observées à maintes reprises chez l'homme et le singe dans des circonstances aussi variées que sous anes- thésie, avec stimulation passive, et lors de tâches nécessitant un comportement (une détection simple de stimuli, par exemple). Ainsi, le rôle fonctionnel joué par ces interactions à une étape du traitement de l'information si précoce demeure une question ouverte. Il a également été démontré que les temps de réaction en réponse à des stimuli audio-somatosensoriels sont facilités par rapport à leurs homologues unisensoriels indépendamment de leur position spatiale. Ce résultat suggère que l'intégration audio- somatosensorielle se produit dans des régions cérébrales possédant des représentations spatiales à large échelle. Cependant, des expériences qui ont exigé un traitement spatial des stimuli ont produits des effets limités à des conditions où les stimuli multisensoriels étaient, alignés dans l'espace ou encore comme pouvant varier selon la partie de corps stimulée. Il n'a pas été établi à ce jour si ces divergences pourraient être dues aux contraintes liées à la tâche et/ou à la nécessité d'un traitement de l'information spatiale.Hypothèse et résultats expérimentauxDans une première étude, nous avons émis l'hypothèse que les interactions audio- somatosensorielles précoces sont pertinentes pour le comportement. En effectuant un partage des temps de réaction par rapport à la médiane d'un sous-ensemble de données comportementales et électroencépha- lographiques, nous avons constaté que les interactions multisensorielles qui se produisent à des latences précoces (entre 40-83ms) sont spécifique aux temps de réaction rapides indiquant une corrélation directe entre ces interactions neuronales précoces et le comportement.Dans une deuxième étude, nous avons émis l'hypothèse que si l'information spatiale devient perti-nente pour la tâche, elle pourrait exercer une influence sur des mesures comportementales de l'intégration audio-somatosensorielles. Dans deux expériences psychophysiques, nous montrons que même si les participants prêtent attention à l'information spatiale, une facilitation de la détection se produit et ce toujours indépendamment de la configuration spatiale des stimuli. Cependant, la performance de discrimination, quantifiée à l'aide d'un index de sensibilité (d') est altérée suite aux essais multisensoriels en général et de manière plus significative pour les essais multisensoriels non-alignés dans l'espace.Dans une troisième étude, nous avons émis l'hypothèse que des améliorations comportementales pourraient différer selon la partie du corps qui est stimulée (la main vs. la nuque). Des résultats préliminaires suggèrent une dissociation possible entre une facilitation comportementale et les potentiels évoqués. Les temps de réactions étaient influencés par la configuration spatiale uniquement dans le cas ou les stimuli somatosensoriels étaient sur la nuque alors que les potentiels évoqués étaient modulés par l'alignement spatial pour les deux types de stimuli somatosensorielles.ConclusionCette thèse apporte des éléments nouveaux concernant le rôle fonctionnel joué par les interactions multisensorielles précoces de bas niveau. En combinant la psychophysique et la neuroimagerie électrique, nous démontrons la pertinence comportementale des ces interactions dans le système humain normal. Par ailleurs, nous montrons que ces interactions précoces sont hermétiques aux influences dites «top-down» sur le traitement spatial suggérant leur occurrence dans des régions cérébrales ayant accès à des représentations spatiales de grande échelle. Nous soulignons enfin des interactions spécifiques entre l'espace auditif et la stimulation somatosensorielle sur différentes parties du corps. Approfondir la connaissance concernant les bases neuronales et les mécanismes impliqués dans l'intégration multisensorielle dans le système normale est d'une importance centrale car elle permettra d'examiner et de mieux comprendre comment le cerveau déficient pourrait bénéficier d'une réhabilitation avec la stimulation multisensorielle.

Looming signals reveal synergistic principles of multisensory integration.

Multisensory interactions are a fundamental feature of brain organization. Principles governing multisensory processing have been established by varying stimulus location, timing and efficacy independently. Determining whether and how such principles operate when stimuli vary dynamically in their perceived distance (as when looming/receding) provides an assay for synergy among the above principles and also means for linking multisensory interactions between rudimentary stimuli with higher-order signals used for communication and motor planning. Human participants indicated movement of looming or receding versus static stimuli that were visual, auditory, or multisensory combinations while 160-channel EEG was recorded. Multivariate EEG analyses and distributed source estimations were performed. Nonlinear interactions between looming signals were observed at early poststimulus latencies (∼75 ms) in analyses of voltage waveforms, global field power, and source estimations. These looming-specific interactions positively correlated with reaction time facilitation, providing direct links between neural and performance metrics of multisensory integration. Statistical analyses of source estimations identified looming-specific interactions within the right claustrum/insula extending inferiorly into the amygdala and also within the bilateral cuneus extending into the inferior and lateral occipital cortices. Multisensory effects common to all conditions, regardless of perceived distance and congruity, followed (∼115 ms) and manifested as faster transition between temporally stable brain networks (vs summed responses to unisensory conditions). We demonstrate the early-latency, synergistic interplay between existing principles of multisensory interactions. Such findings change the manner in which to model multisensory interactions at neural and behavioral/perceptual levels. We also provide neurophysiologic backing for the notion that looming signals receive preferential treatment during perception.

The efficacy of single-trial multisensory memories.

This review article summarizes evidence that multisensory experiences at one point in time have long-lasting effects on subsequent unisensory visual and auditory object recognition. The efficacy of single-trial exposure to task-irrelevant multisensory events is its ability to modulate memory performance and brain activity to unisensory components of these events presented later in time. Object recognition (either visual or auditory) is enhanced if the initial multisensory experience had been semantically congruent and can be impaired if this multisensory pairing was either semantically incongruent or entailed meaningless information in the task-irrelevant modality, when compared to objects encountered exclusively in a unisensory context. Processes active during encoding cannot straightforwardly explain these effects; performance on all initial presentations was indistinguishable despite leading to opposing effects with stimulus repetitions. Brain responses to unisensory stimulus repetitions differ during early processing stages (-100 ms post-stimulus onset) according to whether or not they had been initially paired in a multisensory context. Plus, the network exhibiting differential responses varies according to whether or not memory performance is enhanced or impaired. The collective findings we review indicate that multisensory associations formed via single-trial learning exert influences on later unisensory processing to promote distinct object representations that manifest as differentiable brain networks whose activity is correlated with memory performance. These influences occur incidentally, despite many intervening stimuli, and are distinguishable from the encoding/learning processes during the formation of the multisensory associations. The consequences of multisensory interactions that persist over time to impact memory retrieval and object discrimination.

Learning collaboration in General Internal Medicine: a pilot project for undergraduate students in nursing, physiotherapy and medicine

Background: A form of education called Interprofessional Education (IPE) occurs when two or more professions learn with, from and about each other. The purpose of IPE is to improve collaboration and the quality of care. Today, IPE is considered as a key educational approach for students in the health professions. IPE is highly effective when delivered in active patient care, such as in clinical placements. General internal medicine (GIM) is a core discipline where hospital-based clinical placements are mandatory for students in many health professions. However, few interprofessional (IP) clinical placements in GIM have been implemented. We designed such a placement. Placement design: The placement took place in the Department of Internal Medicine at the CHUV. It involved students from nursing, physiotherapy and medicine. The students were in their last year before graduation. Students formed teams consisting of one student from each profession. Each team worked in the same unit and had to take care of the same patient. The placement lasted three weeks. It included formal IP sessions, the most important being facilitated discussions or "briefings" (3x/w) during which the students discussed patient care and management. Four teams of students eventually took part in this project. Method: We performed a type of evaluation research called formative evaluation. This aimed at (1) understanding the educational experience and (2) assessing the impact of the placement on student learning. We collected quantitative data with pre-post clerkship questionnaires. We also collected qualitative data with two Focus Groups (FG) discussions at the end of the placement. The FG were audiotaped and transcribed. A thematic analysis was then performed. Results: We focused on the qualitative data, since the quantitative data lacked of statistical power due to the small numbers of students (N = 11). Five themes emerged from the FG analysis: (1) Learning of others' roles, (2) Learning collaborative competences, (3) Striking a balance between acquiring one's own professional competences and interprofessional competences, (4) Barriers to apply learnt IP competences in the future and (5) Advantages and disadvantages of IP briefings. Conclusions: Our IP clinical placement in GIM appeared to help students learn other professionals' roles and collaborative skills. Some challenges (e.g. finding the same patient for each team) were identified and will require adjustments.