Tufts Researchers Try to 'Read' Users' Minds

Tufts University researchers have launched a three-year research project aimed at developing methods that would let computers respond to the brain activity of people using the machines.

The effort is funded with a US$445,000 grant from the National Science Foundation.

The project will use light to measure blood flow in the brain, which can be used to identify feelings of work overload, frustration or distraction among computer users, said Robert Jacob, a computer science professor at the Medford, Mass. university. The computer would adjust its user interface based on the measurements of brain activity, he said.

"If the computer knew a little more about you, it could behave better," Jacob said. "If it knew your workload was increasing, maybe it could adjust the layout of the screen. If it knew which air traffic controllers were overloaded, the next incoming plane could be assigned to another controller."

Project team members Jacob and Sergio Fantini, a Tufts biomedical engineering professor, will use functional near-infrared spectroscopy (fNIRS) technology, which uses light to monitor the brain's blood flow to determine stress levels, Jacob said.

While light normally passes through human tissue, it is absorbed when it encounters oxygenated or deoxygenated hemoglobin, Jacob added. Researchers believe that a flow of oxygenated blood to a certain area of the brain happens to replenish blood that is used for taxing tasks.

As part of the project, computer users will wear a futuristic looking head band that shines light on their foreheads. The device will measure what areas of the brain are absorbing the light as the user performs an increasingly difficult task, Jacob said. The information will be relayed to the computer, which will use machine learning techniques to adjust the user interface.

Jacob said that one challenge facing researchers is ensuring that the system makes only gradual changes to the interface to avoid jarring users.

"We're picking up very lightweight subtle information," Jacob said. "We're not always sure we're getting perfect information so we have to respond in a lightweight way. We've got to respond in gentle ways."

The researchers are presenting the results of early tests to determine how user workloads can be monitored using fNIRS this week at the Association for Computing Machinery symposium in Newport, R.I.