Emotion recognition
Emotion recognition is the process of abstraction and classification of audio-visual signals and their decoding as signs of emotions . It is mainly used in emotional psychology , psychotherapy , human ethology and robotics . In robotics, a distinction is made between visual emotion recognition and acoustic emotion recognition . In contrast to human ethology and psychology, robotics has a very simple picture of human emotions. Here the expression and the representation of emotions are often equated with the feeling of emotions. The following example should illustrate this. As soon as a person pulls the corners of his mouth upwards or z. B. speaks the sentence “I am happy”, robotics interprets it as joy. Psychology, on the other hand, would initially only see the expression of a happy sentence - but far from real joy .
Scientific definition of emotion
An emotion must be distinguished from the concept of feeling, mood and personality. A feeling is e.g. B. a shock that one feels when suddenly a masked person appears behind a wall. Then you feel fear . A feeling only becomes an emotion when this physical change is cognitively assessed.
If someone z. B. attributes his palpitations to the masked person, one would speak of fear. However, if he attributes it to his secretly adored one, one would speak of joy. Emotions usually only last a few seconds and have a clearly delimited onset (on-set) and end (off-set). Moods, on the other hand, can last for hours, days or even weeks. If someone says they are in a bad mood today, they are in a bad mood. However, this doesn't necessarily have anything to do with emotions.
Often a certain mood can increase or decrease the likelihood of a certain emotion, but the two things have to be analytically separated. Lastly, a person's personality needs to be demarcated from mood. A choleric person, for example, is permanently negatively overexcited. In this way one can imagine the terms feeling, emotion, mood and personality arranged on a timeline - with feeling on the one hand, short-term, and personality on the other, long-term.
Recognition of emotions in human ethology
The human ethology as a branch of behavioral biology attempts to detect using many components of human emotions. In doing so, she pays particular attention to components of human behavior that have a high information content. In addition to the whole body movement, this includes the sum of changes in movement, gestures, changes in voice frequency and certain parts of facial expressions. The most famous representative of human ethology in Europe is the Institute for Human Ethology in Vienna under the direction of Prof. Karl Grammer.
Cross-race effect
The recognition of emotions between two people is subject to strong fluctuations. A phenomenon called the cross-race effect has been discovered in psychology . This phenomenon says that the emotion detection rate is lower if the emotion to be detected belongs to a face that does not belong to the same culture or ethnicity as that of the observer. However, this effect can be overcome by some form of exercise.
Visual facial expression recognition
This part is commonly called facial expression recognition . A digital video camera or an equivalent optical input device is used as the interface between man and machine. Here, the methods of face recognition are used to analyze the characteristics of the face surface . By Automatic classification , it is possible to assign the facial expression of the serial frames a cluster that could possibly be associated with an emotion. However, studies have shown that only 30% of the emotions expressed through facial expressions also correspond to the emotions that are actually felt. That is why one should not equate visual facial expression recognition with visual emotion recognition. The biological background of visual emotion recognition is the simulation of a human optic nerve in a robot .
Emotion induction
For experimental settings in the field of emotion psychology, behavioral ethology, neuropsychology and many other sciences, it is often important to “generate” emotions in a targeted manner under laboratory conditions . Emotion induction is one of the most difficult areas of emotion research. On the basis of several meta-analyzes on this topic, several methods were extracted with which one can induce emotions most validly.
The first priority is to capture the emotion in reality (keyword field research). Due to poor internal validities , however, this is often not done. The second method, which combines a high internal and a high external validity, is the method of emotional recall, in which an attempt is made to evoke memories from the emotional memory. For experiments outside of EEG emotion research, induction methods such as IAPS or induction methods that allegedly use emotion-inducing film sequences or pieces of music are not recommended. All of these methods fail to provide evidence of specific effectiveness. Robotics often makes use of idealized experimental processes, e.g. B .:
- One induction method is said to induce an emotion in humans.
- Humans express their emotions with a changed face surface.
- A webcam on the computer captures the new facial expression.
- The computer can automatically classify the emotion by classifying it as the emotion that was previously induced.
After completing the learning phase, the AI should be able to recognize emotions independently without being taught this by a human beforehand. Since the induction method is often neither examined for its effectiveness nor the induced emotions are evaluated during the experiment itself, these idealized experimental processes in robotics often remain faulty and incomplete.
Lie detection
Multi-sensory emotion perception is helpful in assessing the truthfulness of utterances, more precisely in recognizing lies, whereby lies are to be understood as deliberately false statements designed to be deceptive. Although no general indicator for the reliable detection of lies can be made out, facial expressions, gestures, language and posture can provide clues. Unconscious or uncontrollable signals such as pupil size, direction of gaze or blushing are relatively reliable. In addition, greater attention should be paid to inconsistencies between the various verbal and non-verbal expressions a person uses.
See also
- Understanding images
- Cross-race effect
- Classification procedure
- Machine vision
- Pattern recognition
- Prosody recognition : purely language-based, automatic classification of suprasegmental features, which among other things can serve to recognize emotions
- Paul Ekman , Facial Action Coding System
literature
- ↑ Study funded by the Federal Ministry of Economics ( Memento of the original from April 17, 2010 in the Internet Archive ) Info: The archive link was inserted automatically and has not yet been checked. Please check the original and archive link according to the instructions and then remove this notice. Reversal of the cross-race effect
- ↑ duden.de (June 5, 2014, 1:13 pm)
- ↑ Litzke, SM (Hrsg.) (2003.) Nonverbal features of lies and power, in: Nachrichtenpsychologie 1. Brühl: University of Applied Sciences of the Federation for Public Administration, Department of Public Security.
- ↑ Lukesch, H. (2003). Recognizability of the lie: everyday theories and empirical findings.
