Our ears reveal our feelings: The path to machines with super-emotional intelligence?

The muscles of the human ear reveal what and how we feel. This is the finding of Dr. Farah Corona-Strauss and Sophie Bonnaire, researchers at the Systems Neuroscience & Neurotechnology Unit in Saarbrücken and Homburg. Their research shows that activity in our ear muscles indicates not only who we are listening to and how attentively, but also the intensity of our emotional response, all within a fraction of a second. The discovery could have important implications for the debate surrounding artificial intelligence, raising the question of what a machine should be permitted to do if it can recognize emotions more accurately than any human.

The work has now been presented in two publications by the team led by Professor Daniel J. Strauss. The research findings appear in the IEEE Transactions on Affective Computing, while a ‘Perspective’ article in Frontiers in Artificial Intelligence introduces and defines the term ‘super-emotional intelligence’.

‘Our ear muscles respond immediately when an individual event moves us emotionally, such as hearing a sound or a face appearing briefly in a scene. This happens within a fraction of a second and it’s not something we can perceive visually,’ says Professor Daniel J. Strauss, explaining his group’s new findings. What makes this particularly remarkable is that in 1871 Charles Darwin described the human ear muscles as vestigial, and a year later, he concluded that humans, unlike animals, had lost the ability to express emotions through their ears. In animals, he described such responses in precise detail: ears flattened in fear or anger and pricked up in attention, as will be familiar to anyone who lives with a dog or cat.

‘We have now been able to show that this evolutionary remnant in our heads, the neural control of the ear muscles, reveals not only who we are listening to but also how we feel while listening,’ says Strauss, a neuroscientist and computer scientist, adding: ‘Although we can no longer point our ears towards a sound, the muscle signal remains. What Darwin believed had been lost is actually still there – it has merely grown quieter.’

Back in 2020, Strauss’s neurotechnology research group reported in the journal eLife that the muscles around the human ear become active whenever a new or significant sound occurs. In those experiments, participants unconsciously made minute ear movements in precisely the direction in which they were focusing their attention. The researchers demonstrated this, among other things, by recording electrical signals from the ear’s vestigial motor system (see the press release ‘Our animal inheritance: Humans perk up their ears, too, when they hear interesting sounds’).  In subsequent studies, the group showed that activity in the ear muscles can also indicate how much effort listening requires, including in people who wear hearing aids.

Laughter leads to a new line of research

During the 2020 experiments, Farah Corona-Strauss noticed that the tiny intrinsic muscles of the outer ear also became active when participants laughed, with the onset of muscle activity occurring shortly before the laughter began. This observation led the neurotechnologist to ask a new question: If the ear responds to sounds that matter to us, does it also respond to things that affect us emotionally? Together with research associate Sophie Bonnaire, she set out to investigate. Bonnaire made the question the subject of her doctoral research at Saarland University and conducted a thorough, systematic examination. The research involved showing participants short film clips, some pleasant, some unpleasant, while surface electrodes recorded activity in their intrinsic auricular muscles. The participants were asked to rate the degree of emotional arousal evoked by each clip. ‘As the emotional arousal elicited by a scene increases, the muscles become more active within a fraction of a second – irrespective of whether the scene is pleasant or unpleasant. And we also find that the muscular response corresponds to the participants’ own reports,’ Sophie Bonnaire explains. What’s particularly striking is the speed with which the muscles respond to an individual stimulus, whether it’s a sound, a face flashing up briefly on screen or, as in the group’s latest research, even a single word encountered while reading.

Could bodily signals pave the way for super-emotional intelligence?

For Daniel Strauss, the discovery that a person’s ear muscles spontaneously and unconsciously reveal their emotional response has important implications for the current debate surrounding artificial intelligence. ‘Until now, debate has focused almost exclusively on “superintelligence”, in other words, on the question of how much more intelligent than us machines might become.’ Strauss believes that a new concept can now be placed alongside it: ‘super-emotional intelligence’ – an attribute ascribed to technical systems capable of interpreting emotions more accurately than any human being, including physiological signals that no human can perceive directly in others. ‘What we definitely do not mean here is a machine that experiences feelings,’ he clarifies. ‘Super-emotional intelligence describes a system that can recognize, understand and respond to a person’s emotions more effectively than trained human experts.’

In a Perspective article published in Frontiers in Artificial Intelligence, Strauss’s doctoral researcher Sophie Bonnaire provides the first definition of Super-Emotional Intelligence (SEI). She describes the capabilities that such a super-emotionally intelligent system would need to display, explains why today’s AI systems are still a long way from achieving them and considers how clearly defined rules could aid the development of technology that works for people rather than against them. ‘Unlike today’s “empathetic” chatbots, which appear compassionate without really monitoring the state of the person they are interacting with, a super-emotionally intelligent system would need to be capable of self-regulation. The SEI system would select the response most likely to help the person reach their desired objective, while also learning from the effects of its own responses,’ Bonnaire explains. 

Care, not control: Safeguards against misuse are essential

Caution is nevertheless essential. ‘If directed towards a different objective, the same technology could become a “hypermanipulative machine”, detecting states that we ourselves are unaware of and steering us towards goals we have never chosen,’ warns Daniel Strauss. The EU AI Act, which prohibits emotion-recognition systems in schools and workplaces, underscores the contemporary relevance of the issue. ‘We argue that human beings must set the objectives and determine what such a machine should and should not be permitted to do. The guiding principle must be care, not control.’

From hearing aids and earbuds to smart glasses

Farah Corona-Strauss and her lab team are already working on a practical technological application. ‘Hearing aids, earbuds and the arms of smart glasses are positioned exactly where the ear-muscle signals originate. The ear is therefore a natural interface between humans and machines. That’s why we are working with hearing-aid manufacturers and international technology partners. An empathetic AI system that detects when listening fatigue sets in or when something triggers an emotional response could adapt accordingly, without requiring either a camera or a questionnaire,’ explains the neurotechnologist.

Background: The research unit

The Systems Neuroscience & Neurotechnology Unit (SNNU) is jointly affiliated with the Faculty of Medicine at Saarland University and the Faculty of Engineering at Saarland University of Applied Sciences (htw saar). Dr. Farah Corona-Strauss heads the unit’s MINDSCAN Lab, where Sophie Bonnaire also conducts her research. The SNNU currently leads a wide range of national and international research projects at the interface of neuroscience, medicine and technology.
Further information: www.snnu.uni-saarland.de

The two publications:

• Sophie Bonnaire, Mayur J. Bhamborae, Daniel J. Strauss, Farah I. Corona-Strauss: Unraveling Affective States from the Auriculomotor System. IEEE Transactions on Affective Computing. DOI: 10.1109/taffc.2026.3717435
• Sophie Bonnaire, Farah I. Corona-Strauss, Daniel J. Strauss: The Rise of Super-Emotional Intelligence (Perspective). Frontiers in Artificial Intelligence. DOI: 10.3389/frai.2026.1975692

Questions can be addressed to:

Prof. Dr. Dr. Daniel J. Strauss
Systems Neuroscience & Neurotechnology Unit (SNNU)
Saarland University, Faculty of Medicine
Tel.: +49 6841-1624090
Email: daniel.strauss@uni-saarland.de

www.snnu.uni-saarland.de

Sophie Bonnaire, the lead author of both publications, is also available for press interviews. Please send interview requests using the contact details shown above.