Can Our Emotions Keep Us Safe on the Internet?

With advancements in AI seemingly making headlines daily, it sometimes feels that our minds may become mated to, and perhaps even overcome by, machines without any warning. But the scientific reality is far less dramatic; while progress is being made in the field of brain computer interfacing (BCI), linking signals from the brain directly to a computer is still a daunting task. Although tracking brainwave activity via electroencephalography (EEG) began in the 1920’s, serious research into BCI began in the 1970s. Nearly 50 years later, noninvasive, effortless brain-to-computer interfacing is likely still many decades away. For now, the state of the art is at the Alexandru Ioan Cuza University of Iasi where researchers have made unique advancements in AI. Their goal: to integrate BCI and computer assistance to help individuals, especially those with disabilities, more efficiently navigate independently through life.

At Iasi, most of the research into BCI has been directed at aiding the severely disabled. They believe that individuals with physical impairments ranging from limited muscle movements to mental conditions such as epilepsy or bipolar disorder could benefit from technological advancements in this field. For instance, researchers believe BCI could allow impaired individuals to gain access to websites or resources that require authentication, like online financial services or Internet-based education and health programs which, at present, may be impossible to access on their own, greatly enhancing their quality of life.

        The technology in development aims to use BCI to aid users with digital security and identity verification, while simultaneously providing an even more unique and valuable tool – emotional authentication. In other words, brainwave analysis will serve as a type of brain activity surveillance, both recognizing the individual and evaluating their mental state. More specifically, researchers are laying the foundations for a system which can assess if users are in an altered state of mind, such as in a manic episode, at the time of attempted access, and will accordingly allow or restrict their request for resources independently of identity authentication. Furthermore, researchers intend to tabulate data over time to develop a detailed profile for each individual; this will allow for a more thorough assessment of their evolving mental states and the subtleties of their neural activity. Such a compiled database will aid gauging the inherent inconsistencies of emotional data, thereby increasing accuracy and ultimately minimizing fraud and misuse of the resources which clients seek.

        Researchers at Iasi have devised a system based on non-invasive BCI, opting for EEG analysis rather than a chip embedded in the nervous system. This has the advantage of not requiring surgical insertion and of bypassing ethical concerns, but it compromises the richness of the data available for analysis because it cannot evaluate signals from subcortical regions such as the amygdala, hippocampus, and thalamus, which would provide a more detailed assessment of emotional states. Instead, electrodes are strategically placed on the scalp in a specific pattern allowing for study of the lateral and medial prefrontal cortex. The configuration of electrodes was designed to provide 20 specific channels of information gathered from corresponding target brain areas that can then be analyzed to provide insight into which subcortical regions are responsible for different signals. By matching the electrical impulses with the type and location of data received, computer processing can deduce each subject’s emotional state, which can then be encrypted and sent to the resource provider for authentication.

        In order to induce select responses that can be measured by the system, researchers rely on an essential component of human emotions: rememoration, or the recreation of a past emotional response based on memory. This phenomenon occurs when individuals are reminded of an original emotional event, and can be synthetically triggered in clinical settings. Data suggesting rememoration is a valuable tool for assessing emotional states was first introduced by a study led by Paul Ekman at the University of California School of Medicine. In their study, researchers compared participants’ facial muscles and the neurons controlling them for induced emotional responses and subsequent rememoration of the original event. Results indicated that the neural and physical responses produced by rememoration mimicked those triggered by corresponding emotions. For patients with mental disorders, measuring the magnitude of rememoration can provide valuable insight into their current mental states – in times of normal brain activity, patients should show reduced emotional responses to trigger events while in altered states rememoration will be amplified.

        In this study, researchers used EEG to evaluate rememoration and emotion in a sample group of five individuals with a median age of 50 in order to establish a baseline during unaltered mental states. Specific neural responses were evaluated by creating a series of triggers tailored to each individual’s personalized emotional experiences, the type of data that would likely be held by resource providers requesting authentication based on identity and mental state. First, a basic profile was created using participant’s responses to ten basic emotions (fear, love, hope, etc.) with words which they associated with each sentiment that reminded them of their own intense related experiences. Additionally, they selected three pictures and chose corresponding words which they did and did not identify with each image. Finally, they provided a password which they could project mentally and use for additional security via comparison of the brain patterns each time the participant visualized it.

Participants were then subjected to five second intervals of exposure to their emotion triggering words or preset pictures and asked to rememorate the experiences linked to each stimulus. Fifteen seconds were allotted between each stimulus and EEG assessments throughout a 16 minute testing period to record brain responses with corresponding centroids, data points which could be used to assess deviations between baseline and elevated emotional responses. Centroids from each emotional channel were selected, and both the most and least active areas were targeted to create mathematical assessments which could be translated into computational codes that could be adapted for security authentication purposes.

        Results were modified with models to compensate for interference from eye and muscular movements and modulation of cardiac rhythm. Additional distortions included combinations of varying alpha and gamma brain activity as well as mere physical exhaustion of subjects stemming from the length of the trial. In order to isolate alpha from gamma fluctuations, each signal was segmented into 62.5 millisecond fragments and mathematically manipulated; researchers also corrected for fatigue by applying algorithms to cancel noise. Combined, these operations helped to create reliable, replicable data points.

        Subsequent analysis worked to establish connections between various stimuli and corresponding cortical patterns. While some of the emotions were not consistently observed in their designated channels, many emotions were effectively displayed, contributing to an emotional blueprint of neural regions. Interestingly, the cortical patterns the researchers observed coincided with brain areas that were previously known to be affiliated with the same emotional responses they had targeted, substantiating the veracity of the paper’s results. Now, the challenge for researchers is to determine the most effective stimulus for each patient using automated patterning systems. This will allow for the stimulation of specific emotions for each channel and produce more effective and accurate responses, eventually leading to a reliable system of EEG-based emotional analysis.  

In practice, this information would be sent to authorized research providers in an encrypted and time stamped message specifying the location of relevant brain stimuli. These messages would be decoded, verified, and processed as the basis for allowing or restricting access, then stored in each individual’s compiled bank of assessments, which would allow for ever increasing accuracy for both individuals and the system.

This type of process has vast implications for the future; could all public locations soon be monitored by emotional assessment as the most secure verification system? Will all of our neural signals be computationally recorded and our every action analyzed by computers? This future may be possible, but such advances are not imminent; for now, BCI technology is far from creating the hybrid man-machine species that is a regular trope of science fiction. Advancements will, however, hopefully be able to improve the lives of many disabled people by allowing them to securely access crucial online information with personalized and reliable systems with the added capacity to evolve with each patients’ needs.


  1. Tulceanu, V., 2012, Comprehensive brainwave authentication using emotional stimuli, 20th
  2. European Signal Processing Conference (EUSIPCO 2012). 1772-76
  3. Vallabhaneni, A., Wang, T., and He, B., 2015, Brain – computer interface, Bioelectric Engineering. 85-121

Related Posts