Show Summary Details

Page of

PRINTED FROM OXFORD HANDBOOKS ONLINE ( © Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use (for details see Privacy Policy and Legal Notice).

date: 12 December 2019

(p. 559) Glossary

(p. 559) Glossary

  • Action units (AUs)

    —visible results from the contraction or relaxation of one or more muscles, used also to describe higher-level concepts in the Facial Action Coding System (Ekman et al., 2002; Petta et al., 2011). (Cited by Ochs, Niewiadomski, & Pelachaud—Chapter 18)

  • Active learning

    — a form of data preselection by the machine for human labeling to reduce manual labeling efforts by “cherry picking.” (Schuller—Chapter 23)

  • Affect detector

    —a model that can infer student affect in real time. (Baker & Ocumpaugh—Chapter 16)

  • Affect elicitation

    —methods used to evoke (or induce) affective responses in individuals. These methods generally involve presenting a stimulus or immersing the subject in a situation to evoke a response from one or more emotion response systems. The nature of the stimulus varies and could include the presentation of images, film, or music; facial expressions or postures; and social or dyadic interactions, among others. (Kory & D'Mello—Chapter 27)

  • Affect-aware learning technology

    —an intelligent learning technology that considers a learner’s affective and cognitive states in its pedagogical decision making. (D'Mello & Graesser—Chapter 31)

  • Affective AutoTutor

    —a natural language intelligent tutoring system that automatically senses and responds to a learner’s confusion, boredom, and frustration by monitoring facial features, body movements, and conversational cues. (D'Mello & Graesser—Chapter 31)

  • Affective body expressions

    —static postures and/or body movements. In the context of their chapter, the authors specifically refer to postures and motions performed or recognized in the context of an affective state or affective dimension. Body expressions may involve the entire body or only part of it, such as affective gait or affective actions such as knocking. (Bianchi-Berthouze & Kleinsmith—Chapter 11)

  • Affective brain-computer interfaces (aBCIs)

    —devices that allow the detection of the affective state of their users based on the neurophysiological activity associated with such states. (Mühl, Heylen & Nijholt—Chapter 15)

  • Affective dimensions

    —dimensions focused on how the world is experienced. Dimensional theorists consider affective states as existing in a continuous, multidimensional space, with the dimensions being bipolar and independent. The primary affective dimensions investigated are valence (levels ranging from pleasure to displeasure), arousal (levels of alertness ranging from calm to excited) dominance/potency (levels of control over an event) (Mehrabian, 1996), and action tendency (the action that one is ready to make in response to the event) (Frijda, 1986). (Cited by Bianchi-Berthouze & Kleinsmith—Chapter 11)

  • Agent

    —a digital representation in a virtual environment controlled by computer algorithms. (Bailey & Bailenson—Chapter 37)

  • Appraisal

    —process whereby people perceive or interpret the evaluative significance of an emotional object or event, typically as an antecedent to reacting emotionally. (Parkinson—Chapter 6)

  • Approach and withdrawal motivations

    —fundamental motivational states on which emotional reactions are based. The approach system controls appetitive and other goal directed behaviors, while the withdrawal system facilitates behavior that removes the individual from sources of aversive stimulation. The left prefrontal cortex plays a key role in approach motivation (including positive affect, social engagement, and anger) while the right prefrontal cortex (e.g., fear) plays a key role in withdrawal. (Kemp, Krygier, & Harmon-Jones—Chapter 4)

  • Autism spectrum disorder (ASD)

    —a pervasive developmental disorder characterized by disordered social communication (the absence or miscoordination of social gazes, facial expressions, gestures, and vocalizations) and clinically salient repetitive behaviors (or unusual interests) that are present by three years of age. (Messinger et al.—Chapter 39)

  • Automated measurement

    —the use of machine learning to map image data to behavioral codes (annotations). (Messinger et al.—Chapter 39)

  • Autonomous

    —functioning independently without explicit control from the outside. (Broek, Janssen, & Westerink—Chapter 35)

  • Avatar

    —a digital representation in a virtual environment controlled by human actions. (Bailey & Bailenson—Chapter 37)

  • Backward masking

    —a method used to block conscious awareness of a visual stimulus. The target stimulus is shown to an individual very briefly (e.g., 15 to 60 milliseconds), followed immediately by a “mask” stimulus shown for a longer time (e.g., 500 milliseconds). Individuals report being consciously aware of only the mask. (Kory & D'Mello—Chapter 27)

  • Behavior manipulation

    —a method in which individuals are instructed to adopt particular behaviors—such as body postures or facial expressions—in order to induce particular affective states. (Kory & D'Mello—Chapter 27)

  • Behavior Markup Language

    —representation language comprising all those representations that are necessary for the realization of behavior. It includes directives for the realization of textual and prosodic information, facial display, gestures and postures, eye gaze, and, very importantly, directives for the temporal synchronization of behaviors (Petta et al., 2011). (Cited by Ochs, Niewiadomski, & Pelachaud——Chapter 18)

  • Biofeedback

    —the use of measurements of physiological functions in order to control them. (Broek, Janssen, & Westerink—Chapter 35)

  • Brain-computer interfaces (BCIs)

    —mechanisms that allow for the control of devices and applications based on the neurophysiological activity of a user, thereby bypassing muscular pathways. (Mühl, Heylen, & Nijholt—Chapter 15)

  • (p. 560) Closed-loop model

    —control systems with an active feedback loop. (Broek, Janssen, & Westerink—Chapter 35)

  • Collaborative virtual environment (CVE)

    —a virtual environment that supports multiple users from remote locations in a common virtual space. (Bailey & Bailenson—Chapter 37)

  • ConfusionTutor

    —a learning environment that aims to promote deeper comprehension by strategically induces confusion in the minds of learners. (D'Mello & Graesser—Chapter 31)

  • Contextual features

    —features hypothesized or discovered to play key roles of relevance in interpreting unfolding events and states beyond an isolated occurrence, taking into account the context of interaction (e.g., task, user preferences, presence of other people, behavior of the interactants, etc.). The consideration of context is necessary in cases where the meaning of a feature of interest cannot be determined or disambiguated in isolation (i.e., without reference to other features). (Castellano, Gunes, Peters, & Schuller—Chapter 17)

  • Crystal Island

    —a learning technology that embeds the learning content in a narrative-centered game supporting narrativity, realism, and immersion. (D'Mello & Graesser—Chapter 31)

  • Cultural dichotomies

    —Scales used to characterize different cultures by their position between two poles. Examples of culture dichotomies include power distance, identity, gender, uncertainty avoidance, long-term orientation, and context. (Andre—Chapter 22)

  • Cyberpsychology

    —a new branch of psychology that aims at the understanding, forecasting, and induction of the different processes of change related to the use of new technologies. (Riva, Calvo, & Lisetti—Chapter 41)

  • Cybertherapy

    —the branch of cyberpsychology that tries to understand how technologies can be used to induce clinical change. (Riva, Calvo, & Lisetti—Chapter 41)

  • Discrete emotions

    — instances of unique and separate states (e.g., anger or happiness). Many discrete emotion theorists also consider a number of emotions as basic or primary, yet there is no consensus on either the number of categories or which emotions are considered basic (Ortony & Turner, 1990). (Cited by Bianchi-Berthouze & Kleinsmith—Chapter 11)

  • Display rule

    —cultural norm about when, where, and with whom it is appropriate to express or not express a particular emotion on the face. (Parkinson—Chapter 6)

  • Dyadic interaction

    —a social interaction specifically between two individuals (see Social Interaction). One affect elicitation method focuses on bringing pairs of individuals together to engage in an unrehearsed, minimally structured conversation in order to evoke affective states in a more naturalistic context. (Kory & D'Mello—Chapter 27)

  • Educational data mining

    —the research area that uses data mining methods to model and understand learners and learning. Closely related to learning analytics. (Baker & Ocumpaugh——Chapter 16)

  • Electroencephalography (EEG)

    —a portable neuroimaging method for the temporally high-resolution recording of variations in electrophysiological brain activity from the scalp. (Mühl, Heylen, & Nijholt—Chapter 15)

  • Embodied conversational agent (ECA)

    —a humanlike conversational character able to engage with the user in multimodal communication. The usual modalities include speech, facial expression, eye gaze, head movement, body posture, and hand-arm gesture (Petta et al., 2011). (Cited by Ochs, Niewiadomski, & Pelachaud—Chapter 18) (Messinger et al.—Chapter 39)

  • Emotion

    —a term used colloquially to reflect a wide range of affective responses (feelings, mood, disposition, etc.). In emotion theory, emotions are generally perceived as short-term affective responses and often perceived as both “basic” versus “social/moral/higher-order” emotions, where the first category is more often tied to primary physiological responses. (Healey—Chapter 14). Often defined also as a multicomponent response to a significant stimulus characterized by brain and bodily arousal and a subjective feeling state that elicits a tendency toward motivated action. Note, however, that there may be instances of emotion in which significant stimuli (e.g., emotions without obvious causes), subjective feeling states (e.g., unconscious emotions), and motivated action (e.g., sadness) are not necessary. (Kemp, Krygier, & Harmon-Jones—Chapter 4)

  • Emotion contagion

    —process whereby an individual (automatically) “catches” the emotional state of another individual, often thought to be mediated by mimicry of expressive movements or generation of complementary motor codes. (Parkinson—Chapter 6)

  • Emotion object

    —what an emotion is about. The (intentional) “object” may be an imagined or anticipated event rather than a physical thing. (Parkinson—Chapter 6)

  • Emotion regulation

    —the process of actively modifying the causes, content, or consequences of emotion. (Parkinson—Chapter 6)

  • Emotion theories

    —while there are many different theories of emotion, a significant proportion of affective computing research focuses on discrete emotions and/or affective dimensions. (Bianchi-Berthouze and Kleinsmith—Chapter 11)

  • Emotional film clips

    —short movie segments, usually including both images and sound, that have been selected and evaluated for their potential to evoke affective states in the viewer. (Kory and D'Mello—Chapter 27)

  • Emotional images

    —digital images or photographs that have been carefully selected and evaluated for their potential to evoke affective states in the viewer. (Kory & D'Mello—Chapter 27)

  • Emotional music

    —a recorded musical piece that has been selected and evaluated for its ability to evoke affective states in the listener. (Kory & D'Mello—Chapter 27)

  • Empathy

    —the feeling of being affected by other people’s emotions because you care about those people or see things from their perspective. (Parkinson—Chapter 6)

  • Evaluator

    —a term used synonymously with annotator, labeler, and rater in this section for the person who attaches labels to affective data. (Schuller—Chapter 23)

  • Event-related potentials (ERPs)

    —a stereotyped electrophysiological response to a specific stimulus or event that is estimated by averaging the recorded EEG traces recorded immediately after several occurrences of the same event. (Mühl, Heylen, & Nijholt—Chapter 15)

  • Facial action coding system

    —a categorization system for facial behaviors based on the underlying musculature. Facial behaviors are coded in terms of action units involved in a change in appearance as well as duration, intensity, and asymmetry (Petta et al., 2011). (Cited by Ochs, Niewiadomski, & Pelachaud—Chapter 18)

  • Feature representation and selection

    —descriptions of the way features are represented to a machine-learning algorithm. Usually, this is done with a vector or time series of (p. 561) vectors. The selection of features deals with reaching a more compact representation—usually by excluding features of lower relevance. This is commonly done either by a suited measure such as correlation or information gain or with the classifier in the loop. In addition, a search function that leads to a local optimum rather than a global one is needed mainly for efficiency reasons. (Castellano, Gunes, Peters, & Schuller—Chapter 17)

  • Features

    —attributes of a data record used as predictors in a data mining analysis. (Baker & Ocumpaugh—Chapter 16)

  • Functional magnetic resonance imaging (fMRI)

    —a neuroimaging method for the spatial high-resolution recording of brain activity by detecting associated changes in blood flow. (Mühl, Heylen, & Nijholt—Chapter 15)

  • Functional near-infrared spectroscopy (fNIRS)

    —a portable neuroimaging method for the recording of brain activity by detecting associated changes in blood flow via magnetic impulses. (Mühl, Heylen, & Nijholt—Chapter 15)

  • Galvanic skin response

    —also commonly referred to as electrodermal activity (EDR), or skin conductance, this is a commonly used physiological metric that determines a person’s sweat levels by measuring the conductance of the skin. The skin is normally an insulator, but sweat is ionic and conducts electricity, so that when a person starts sweating, skin conductivity increases. This phenomenon is most often measured by placing two electrodes on two adjacent fingers and measuring the voltage in response to a small injection current that runs between the two electrodes across the skin of the palm of the hand, where many of the most emotionally reactive sweat glands are found. (Healey—Chapter 14)

  • GazeTutor

    —a learning environment that senses and responds to patterns of disengagement by monitoring eye gaze. (D'Mello & Graesser—Chapter 31)

  • Gold standard

    —represents the compromise made to get as close as possible to the ground truth if the phenomenon cannot be easily measured. For affective data, this may be difficult to reach, and several evaluators (see below) are often used to get closer to the ground truth. (Schuller—Chapter 23)

  • Ground truth

    —the affective label assigned to an affective expression. This ground truth affective state or dimension level may be predetermined by the researcher or assigned by expert or nonexpert observers (i.e., people who judge the affective state or level of affective dimension by viewing or listening to the affective expression). (Bianchi-Berthouze & Kleinsmith—Chapter 11)

  • Heart rate variability

    —a term used to describe how successive heart beats differ from one another (e.g., how the lengths of the intervals between successive heart beats vary). The term heart rate variability is used to describe a number of metrics, some of which are calculated in the time domain and others in the frequency domain. (Healey—Chapter 14)

  • High-risk sibling

    —the younger brother or sister of a child with an autism spectrum disorder (ASD). Typically studied before three years of age, these siblings are themselves at risk both for an ASD and ASD-related symptoms that do not meet criteria for an ASD diagnosis. (Messinger et al.—Chapter 39)

  • Immersive virtual environment technology (IVET)

    —technology that immerses users in a sensory rich virtual environment (e.g., onethat provides visual, haptic, and olfactory feedback). (Bailey & Bailenson—Chapter 37)

  • Infant-parent interaction

    —the exchange of communicative signals between infants (typically twelve months of age and under) and parents in whom the behavior of infant or parent (or both) influences the behavior of the other partner; also referred to as emotional communication. (Messinger et al.—Chapter 39)

  • Intelligent Tutoring System

    —An online learning system that provides interactive activities and adapts in real time to differences in student learning, behavior, affect, or other individual differences. (Baker & Ocumpaugh—Chapter 16)

  • Interaction log

    —a detailed historical record of behavior enacted within a computerized learning system by one or more students. It typically includes data on the student actions, the system responses, and any semantic interpretation of student behavior that is feasible at run time. By definition, interaction logs do not include data from behavioral and physiological sensors. (Baker & Ocumpaugh—Chapter 16)

  • Interpersonal emotion transfer

    —phenomenon of one person’s emotion inducing a corresponding emotion in someone else (operating by a range of processes). (Parkinson—Chapter 6)

  • Lexicon

    —a list of correspondences between signals and meanings. (Poggi, Pelachaud, & de Rosis, 2000). (Cited by Ochs, Niewiadomski, & Pelachaud—Chapter 18)

  • Magnetoencephalography (MEG)

    —a neuroimaging method for the temporally high-resolved recording of variations in electric brain activity by detecting associated changes in the magnetic fields. (Mühl, Heylen & Nijholt—Chapter 15)

  • Mimicry

    —making movements whose characteristics correspond to the characteristics of observed movements being made by other people. (Parkinson—Chapter 6)

  • Mirror neuron

    —nerve cell (or group of cells) that fires both when a movement is observed in someone else and when the same movement is enacted by self. (Parkinson—Chapter 6)

  • Modality fusion

    —An approach to combine data at different levels, from multiple homogenous or heterogeneous streams, to predict the final label or class. (Castellano, Gunes, Peters, & Schuller—Chapter 17)

  • Modeling

    —quantitative characterization of relationships between the components of complex expressive or communicative systems. (Messinger et al.—Chapter 39)

  • Mood

    —a relatively long-lasting emotional state. (Broek, Janssen, & Westerink—Chapter 35)

  • Multimodal

    —more than a single modality is present. In affective databases, typically such modalities include audio, video, physiological, and textual data. Different views exist on what is multimodal, or rather multistream: For example, speech contains acoustic and usually verbal content as well. Strictly speaking, the modality is speech, but often the combination of these two streams—acoustic and textual content—is already considered multimodal. (Schuller—Chapter 23)

  • Multimodal affect recognition

    —a process that performs automatic affect recognition by using several input modalities, such as behavioral (e.g., face, gesture, posture, speech prosody, etc.,), physiological (e.g., electrodermal activity, etc.), and contextual (e.g., task, user preferences, etc.) data. (Castellano, Gunes, Peters, & Schuller—Chapter 17)

  • Music

    —(the product of) an art form deploying sound, silence, rhythm, melody, etc. (Broek, Janssen, & Westerink—Chapter 35)

  • Natural kinds

    —fundamental processes in the brain that exists across species and human cultures; a phenomenon that is discovered, not created, by the human mind. In this regard the basic emotions are characterized as “natural kinds,” hardwired into the brain and associated with distinctive patterns of neural activation. Note that different conceptualizations of the basic (p. 562) emotions have been proposed (e.g., Ekman versus Panksepp). (Kemp, Krygier, & Harmon-Jones—Chapter 4)

  • Nonverbal behavior

    — behavior corresponding to “facial expressions, body language, social touching, vocal acoustics, and interpersonal distance” (Ambady & Weisbuch, 2010). Nonverbal behavior may convey several kinds of information—for instance, on one’s emotions or attitude. Nonverbal communication “refers to the sending or the receiving of thoughts and feeling via nonverbal behavior” (Ambady & Weisbuch, 2010). (Cited by Ochs, Niewiadomski, & Pelachaud—Chapter 18)

  • Overfitting

    —a model is said to be overfitted when it so closely matches the data it was trained on that it cannot generalize to new data, such as data from new students or different populations. Overfitting typically occurs as a result of using an overly flexible or complex model for the data set size and the actual strength of the relationship(s) being modeled. An assessment of model goodness that is less vulnerable to overfit can be obtained through the use of appropriate cross-validation, with student-level cross-validation considered particularly useful in educational domains. (Baker & Ocumpaugh—Chapter 16)

  • Personalization

    —the process of accommodating the differences between individuals. (Broek, Janssen, & Westerink—Chapter 35)

  • Physical presence

    —the perceptual experience that measures how real the virtual environment and the objects within it feel. (Bailey & Bailenson—Chapter 37)

  • Physiological

    —in general, physiology is a branch of biology that deals with the functions of activities of life. With respect to affective computing, physiological affect in general refers to responses that come from the body, more especially those associated with the autonomic nervous systems. Although brain activity is in essence physiological, the field of neurophysiology provides a more specific view of brain function and the term physiological is usually used to refer to other types of bodily responses. (Healey—Chapter 14)

  • Positive computing

    the design and development of technology to support well-being and human potential (Riva, Calvo, & Lisetti—Chapter 41)

  • Positive technology

    —the branch of cyberpsychology that uses technology to manipulate and enhance the features of our personal experience for increasing wellness and generating strength and resilience in individuals, organizations, and society. (Riva, Calvo, & Lisetti—Chapter 41)

  • Positron emission tomography (PET)

    —a neuroimaging method for the spatial high-resolution recording of brain activity by detecting associated changes in blood flow via radioactive tracers. (Mühl, Heylen, & Nijholt—Chapter 15)

  • Presence

    —the subjective psychological experience of being in a virtual environment. (Bailey & Bailenson—Chapter 37)

  • Proactive systems

    —an affect-aware learning technology that aims to induce or impede certain affective states. (D'Mello, Graesser—Chapter 31)

  • Psychological constructionism

    —this view considers emotions as a construct resulting from more basic building blocks such as core dimensions like approach-withdrawal or valence-arousal. The debate over whether emotions are “natural kinds” versus a “psychological construction” has been likened to the Hundred Years’ War between England and France (Lindquist, Siegel, Quigley, & Barrett, 2013). (Cited by Kemp, Krygier, & Harmon-Jones—Chapter 4)

  • Psychophysiology

    —the field that studies the impact of psychological states on the physiological system and vice versa. (Broek, Janssen, & Westerink—Chapter 35)

  • Reactive systems

    —an affect-aware learning technology that senses and responds to affective states. (D'Mello & Graesser—Chapter 31)

  • Reappraisal

    —the process whereby the perceived emotional meaning of an event is modified. May be actively used as a means of emotion regulation. (Parkinson—Chapter 6)

  • Resource

    —a term used here as a synonym for corpus or database. (Schuller—Chapter 23)

  • Robot

    —a hardware-based agent with sufficient autonomy to interact with children—for example, to assist with emotion or other learning tasks. (Messinger et al.—Chapter 39)

  • Self presence

    —the extent to which a person identifies with how he or she is digitally represented in the virtual environment or the level in which the virtual self is experienced as the actual self. (Bailey & Bailenson—Chapter 37)

  • Sensing

    —using a physical instrument to detect a physical stimulus. In affective computing, sensing is used to capture information that can be used by a computer to incorporate into algorithms, for example to sense skin conductance, a GSR sensor is used and to sense heart rate a heart rate sensor (such a an electrocardiogram –ECG) is used. (Healey—Chapter 14)

  • Signals

    — a time varying response that communicates information about phenomena. In the context of physiological affective computing, a signal is usually a two-dimensional time-voltage signal, measured from some part of the body. For example, a skin conductance signal conveys information about how a person’s sweat level changes over time and a heart rate signal conveys information about how a person’s heart rate changes over time. (Healey—Chapter 14)

  • Smart health

    —a new branch of medicine that uses the latest technological advances (e.g., sensors and sensors networks, actuators, robots and virtual assistants) to build intelligent care (e.g., smart homes for independent living, wearable prosthetics, lifestyle modification coaching). (Riva, Calvo, & Lisetti—Chapter 41)

  • Social appraisal

    —process whereby another person’s perceived emotion, expression, or behavior modifies one’s appraisal of an emotion object. (Parkinson—Chapter 6)

  • Social attitudes

    —positive or negative evaluation of a person or a group of people. Social attitudes include cognitive elements like beliefs, opinions, and social emotions. (Pantic & Vinciarelli—Chapter 7)

  • Social emotions

    —emotions such as admiration, envy, and compassion that can be felt only toward another person (Pantic & Vinciarelli—Chapter 7)

  • Social evaluations

    —social evaluations relate to assessing whether and how much the characteristics of a person comply with our standards of beauty, intelligence, strength, justice, altruism, etc. (Pantic & Vinciarelli—Chapter 7)

  • Social interaction

    —a relationship between two or more individuals, fleeting or enduring, in which an individual’s actions and behavior are responsive to the actions and behavior of the other or others. In one affect elicitation method, researchers try to create realistic social interaction scenarios that might evoke emotions in a more naturalistic context. (Kory & D'Mello—Chapter 27)

    Social interactions are events in which actually or virtually present agents exchange an array of social actions (i.e., communicative and informative signals performed by (p. 563) one agent in relation to one or more other agents). (Pantic & Vinciarelli—Chapter 7)

  • Social presence

    —the psychological state that measures the extent that other virtual social actors are experienced as actual social actors. (Bailey & Bailenson—Chapter 37)

  • Social relations

    —a social relation is a relation between two (or more) persons in which these persons have related goals. (Pantic & Vinciarelli—Chapter 7)

  • Social signals

    —communicative or informative signals which provide information about social facts (social interactions, social emotions, social evaluations, social attitudes and social relations) (Pantic & Vinciarelli—Chapter 7)

  • Stereotypical expression

    —according to many theorists, there are universal facial expression patterns linked to the six basic emotions (joy, disgust, anger, surprise, sadness, and fear) as defined by Paul Ekman (Ekman and Friesen, 1975). (Cited by Ochs, Niewiadomski, & Pelachaud—Chapter 18)

  • Suppression

    —consequence-focused emotion regulation that reduces visible expression. (Parkinson—Chapter 6)

  • Transformed social interaction (TSI)

    —decoupling of an avatar’s appearance or behavior from the actual person through the use of computer algorithms. (Bailey & Bailenson—Chapter 37)

  • UNC-ITSpoke

    —a speech-enabled intelligent tutoring system that automatically senses and responds to a learner’s uncertainty and response accuracy. (D'Mello & Graesser—Chapter 31)

  • Validation

    —confirming that a product or service meets specifications. (Broek, Janssen, & Westerink—Chapter 35)

  • Weakly supervised learning

    —subsumes different types of machine learning where full supervision is not given. Usually, this means that data without labels are used by the machine to autonomously adapt or (even entirely) train itself by semisupervised or unsupervised learning. (Schuller—Chapter 23)