Can a Computer Read Your Mind?
Photo Source: PxHere
Our minds remain the ultimate private place, a place to ourselves, our thoughts, hopes, dreams, and fears. Nobody can see inside your mind. Such opacity drives people to say “What’s on your mind? Or even offer to pay when they ask, “A penny for your thoughts?” Many science fiction and fantasy stories such as Scanners, Carrie, and Philip K. Dick’s Ubik include mind reading as a plot device. In Star Trek: The Next Generation, a race of humanoids called Betazoids, such as the ship’s counsel Dianna Troi, possess empathic abilities that allow them to read other people’s emotions. Such capabilities prove useful in many situations but can pose a threat when used by enemies. For research purposes, scientists today continue to pursue means of looking into people’s minds with different technologies.
One technology powered by artificial intelligence demonstrated an unprecedented glimpse into the human mind. Functional MRI affords a unique and exciting glimpse into the brain without having to ask the question, “What is on your mind?” Recently, researchers at Purdue University in West Lafayette, Indiana published a series of articles describing work they performed using artificial intelligence and functional magnetic resonance imaging or fMRI (Scientific Reports, 2018). Many of us or a friend or relative has had an MRI scan to test for things like tissue damage to a knee or to assess the effects of a stroke. MRI uses strong magnets, much stronger than your typical fridge magnet, radio waves, and sensitive detectors to observe fine details in soft tissues such as the spinal cord and the brain. The magnetic waves and radio waves in the MRI work together to vibrate water molecules in the body. Sensitive detectors can see how water vibrates in different parts of the body depending on the surrounding tissue, which then builds a detailed, three-dimensional picture of soft tissues in the body. Brain fMRI visualizes blood flow to different parts of the brain and scans taken over time demonstrate how blood flow changes in response to a different stimulus. The researchers at Purdue led by Zhongming Liu studied how blood flow in the brain changes in response to test subjects being shown different pictures in eighty different categories from peoples’ faces to animals, objects, and scenes such beaches and mountain views. Additionally, as they collected the data of how the blood flow changed in parts of the brain depending on whether the test subjects saw a picture of an eagle or an airplane, an artificial intelligence computer learned the patterns of blood flow depending on the different labeled pictures.
After gathering the data from the test subjects, researchers began training an artificial intelligence program to guess what the brain was seeing using the MRI images. Remarkably, after training the AI with known images, the computer, without knowing what the test subjects were looking at, could predict what each picture contained based on the brain’s blood flow. The computer could determine if the test subject was seeing a human face, a bird, an airplane, or a mountain range. Moreover, the AI could predict what pattern of blood flow a new picture would produce and make a shape of it on the computer screen. The images predicted by the AI look like shimmery blobs on the computer screen, not a crisp reproduction, but they had meaning for the computer. The AI obtained the highest accuracy with shapes, but the accuracy declined for more complex images.
The private domain of our thoughts remains our only genuinely autonomous domain. Science fiction and fantasy use mind readers to drive story lines aiding protagonists but more often giving an advantage to villains. Notably, Lord Voldemort in the Harry Potter series used mind reading and manipulation to evil effect. Recently, neuroscience aided by fMRI and artificial intelligence developed a technique to predict what a test subject sees in a picture based on complex changes in blood flow to different parts of the brain. The brain produces specific signatures of blood flow for different images, offering for the first time a glimpse, albeit a foggy glimpse, into the human mind. The results did not return clear pictures of what the test subjects saw in pictures, and moreover, the results did not transfer well from one test subject to the other. However, as we move towards more AI driven techniques to look into the human brain, we must in parallel consider how we protect the final domain of privacy—our thoughts and feelings.
Dr. Smith’s career in scientific and information research spans the areas of bioinformatics, artificial intelligence, toxicology, and chemistry. He has published a number of peer-reviewed scientific papers. He has worked over the past seventeen years developing advanced analytics, machine learning, and knowledge management tools to enable research and support high level decision making. Tim completed his Ph.D. in Toxicology at Cornell University and a Bachelor of Science in chemistry from the University of Washington.
You can buy his book on Amazon in paperback here and in kindle format here.