• Login
  • Register

Work for a Member organization and need a Member Portal account? Register here with your official email address.

Project

NeuroChat

Copyright

Sunday Ajak

Sunday Ajak

Project Contact:
Groups

👉 Read the full paper here (open access).  

NeuroChat is a neuroadaptive AI tutor that combines generative AI with real-time EEG data to create a personalized, dynamic learning experience. Unlike traditional conversational AI, which relies on explicit user input, NeuroChat infers engagement levels directly from brain activity, modifying its responses to match the user's attention level, cognitive state, and learning preferences. When engagement dips, NeuroChat modifies its conversational strategy, adjusting response complexity, pacing, and level of interactivity to re-engage the learner and maintain an optimal cognitive state.

This project addresses one of the most pressing challenges in digital learning and AI-human interaction: how can an AI tutor move beyond static responses and actually sense user engagement? With AI-powered education on the rise, LLM-based tutoring systems are widely used; yet they remain passive, requiring learners to communicate their engagement via prompt engineering manually. With NeuroChat, the system automatically adjusts, reducing cognitive effort and making interaction more fluid and human-like.

Designed for consumer-grade EEG headbands with dry electrodes, with all signal processing occuring in a standard web browser, NeuroChat enables neuroadaptive AI tutoring accessible beyond research environments. In our evaluation, NeuroChat increased cognitive and perceived engagement, as measured by EEG-based engagement indices and user evaluations; however, improvements in immediate learning performance were not observed. Those who preferred conversational, exploratory learning found it especially effective.

Copyright

Dünya Baradari

Beyond education, NeuroChat has implications for self-directed learners, knowledge workers, and accessibility applications. NeuroChat envisions a world where AI assistants don’t just respond to what we say, but also to how we think and feel in the moment.

We are interested in partnering with learning platforms to bridge the gap to enable generative AI to deliver truly personalized and customized support. 

If you are interested in learning more about this project, email nkosmyna AT mit DOT edu  or fill out this form here.

Frequently Asked Questions

  1. How does the brain sensing technology work?

    NeuroChat uses the Muse EEG (electroencephalography) headband to measure your brain activity. The headband has sensors that detect electrical signals from your brain. These signals are processed to extract metrics related to engagement. The system then adjusts the AI tutor's responses based on these metrics, providing more challenging content when you're engaged and simplifying explanations when you're struggling.

  2. Can NeuroChat read thoughts?

    No,  NeuroChat cannot read thoughts. We calculate engagement based on general brain signal patterns, specifically, alpha, beta, and theta band powers.

  3. Can NeuroChat stimulate the brain?

    No, NeuroChat cannot stimulate the brain. The EEG device used can only record general brain signal patterns.

  4. Do you use NeuroChat for children in classrooms?

    NeuroChat has not been tested in a classroom environment and is currently not intended for use by children and teenagers under the age of 18.

  5. What metric is NeuroChat using and have you tested it with other metrics?

    We have tested NeuroChat with the engagement index, which is calculated as the ratio beta/(alpha + theta) band powers. We are currently piloting additional cognitive metrics, including attention and prefrontal alpha asymmetry. 

  6. Which EEG headsets are supported with NeuroChat?

    Currently, we only support the Muse EEG headset. If you are interested in adding support for another EEG device (e.g. Neurable, OpenBCI, Emotiv), please reach out to us here:  https://tally.so/r/wvzeMQ

Research Topics
#human-computer interaction #artificial intelligence #cognition #consumer electronics #learning + teaching