Search

Computer systems, that support learning processes, can adapt to the needs and states of a learner. The adaptation might directly address the knowledge deficits and most tutoring systems apply an adaptable learning path of that kind. Apart from a preliminary knowledge state, there are more factors, that influence education effectiveness and among those there are fluctuating emotional states. The tutoring systems may recognize or predict affective states of a learner and react to them to foster success of an educational process. This paper explores, how techniques derived from affective computing and user-centric design might be applied in intelligent tutoring systems development. An affective intervention is a modification of a standard control path or system behaviour that is a response to user's affective state and aims at providing effective execution of a task (learning). There are several criteria for a good intervention model for an application. A system should make an affective intervention only when required and refrain from intervention otherwise. Interventions should be a natural element of an interaction with application and should be tailored to a user state. The ten step process, called AFFINT, was proposed to design appropriate affective interventions. In this paper the process is adapted to the domain of technologybased learning. The steps of the process are analyzed, providing insight into designing educational systems that react to the recognized or predicted emotional state of a learner. The study is based both on literature review as well as it is supported with a design case study of an educational tool. The paper shows, that combining affective computing and educational systems is possible and beneficial. The paper might be interesting both for researchers, working on educational processes and supportive tools as well as for developers, aiming at designing effective tutoring systems. [ABSTRACT FROM AUTHOR]

In the last decade, the interest in the development of tools and devices for recognizing human emotions in learning process has increased continuously. It was proved that the electrical brain activity using electroencephalography (EEG) represents a useful methodological tool in understanding cortical processes that underlie performance and students' engagement in learning activities. Specific emotional states, like mental stress, concentration, relaxation, fatigue, and cognitive increase activation in Delta (0.5-4 Hz), Theta (4-7 Hz), Alpha 1 and 2 (8-12 Hz), Beta 1 and 2 (12-30 Hz), Gamma (30-70 Hz) frequencies. For example, the increase of frontal Beta-1 spectral power is associated with cognitive tasks demands and the decrease of Beta-1 power values reflects relaxation. Alpha is the dominant frequency in the human EEG and is generated in widespread areas of the cortex through cortico-cortical and thalamo-cortical interactions reflecting emotions. The EEG-based Brain Computer Interface (BCI) systems have been widely studied in different medical labs with high quality EEG recording devices which are much too expensive and need special employment. As an alternative to these professional equipment, several low-cost EEG devices were developed for out of the lab applications e.g. schools, universities, sports medicine, psychology or even neurophysiology. Two pertinent and detailed studies developed at Princeton University (http://compmem.princeton.edu/experimenter) and University of Mons (www.biomedical-engineering-online. com/content/12/1/56) reveal that the Emotiv Epoc headset could be taken into account. This low-cost EEG device has higher relative operational and maintenance costs than its medical-grade competitor, it "is able to record EEG data in a satisfying manner" but it should only be chosen for non critical applications such as games, communication or feedback evaluation in a well-known scenarios. Our study aims to use the Emotiv headset for evaluating the students' emotional state in learning process. For this goal, some steps were completed: EEG data acquisition and analysis. The headset connects wirelessly to any PC via a USB dongle allowing freedom of movement, has 14 channels/sensors and the data sample rate is 128Hz. Due to the problem of eye/head movements automatic artefact detection must be completed and only artefact-free segments must be used for analysis. A good option is to use the Matlab toolbox EEGLAB, freely available open source research software (http://sccn.ucsd.edu/eeglab) and the Independent Component Analysis (ICA) method for general EEG analysis and artifacting. Spectral power for each frequency band is computed and used for statistical analysis. Statistical analysis. Mean values and standard deviation could be calculated and relevant statistical values tests (Kolmogorov-Smirnov, ANOVA or Mauchley) have to be performed for different learning status, like reading a very simple text or a very difficult one, discussion on a subject, playing a game, programming under time restrictions. The obtained row data of the acquired signal should be compared with those provided by MIT-BIH Database Distribution (http://ecg.mit.edu/). Based on this EEG feedback approach, some conclusions can be formulated regarding students' implication in learning or the difficulty of the teaching materials. [ABSTRACT FROM AUTHOR]

In 2004 a group of affective computing researchers proclaimed a manifesto of affective learning that outlined the prospects and white spots of research at that time. Ten years passed by and affective computing developed many methods and tools for tracking human emotional states as well as models for affective systems construction. There are multiple examples of affective methods applications in Intelligent Tutoring Systems (ITS). This paper revises whether the white spots from the 2004 manifesto have been covered as well as explores the progress in affective tutoring systems construction. The article provides also a brief comparison of selected affective tutoring systems. When reviewing affective computing literature in 2014 one might be impressed - there are algorithms that recognize affect from diverse input channels with accuracies reaching over 90%. There are emotional virtual characters and some of them are used in e-learning environments. However, although there is much progress in affective computing research, it seems that more white spots are found than covered, which is typical for relatively young scientific domains. Highest emotion recognition accuracies are obtained for distinguishing two emotions, which is obviously not enough for learning systems. Some of the future challenges include: improvement of both accuracy and granularity of emotion recognition, methods for emotion representation models mapping and comparison, integration of emotion recognition results from multiple input channels, effective intervention models for learning, design patterns and frameworks for affective tutoring systems, models for quantifying and integration of uncertainty related to emotional states recognition, affect-adaptive control flows and more. The thesis of this paper can be summarized as follows: Affective computing grew up from infancy, however it is still far from maturity especially when applied to learning support. During the decade of diverse investigations, affective-cognitive imbalance in ITS has changed in research, however has not changed in learning support tools. [ABSTRACT FROM AUTHOR]

communication apprehension, communication competency, nonverbal immediacy, affective learning, cognitive learning, behavioral outcomes, communibiological paradigm, neuropsychological, temperament, activation threshold, culture, trait-based, treatment ..PAT.-Unpublished Manuscript [ABSTRACT FROM AUTHOR]