I can't seem to respond on your page, so I'll respond here.
I am not exactly sure what you mean by "reactions?" Reactions to what? Emotions? A visual scene? A moving ball? A break up? What they're thinking about?
As of right now, we can "mind read" very basic reactions. I'll use the visual system as an example. There is a primary visual center in the brain. Before reaching this center, the visual information we see are just pieces. We have lines and edges, color, motion, etc. The brain pieces this information together in the visual cortex (oversimplified statement).
Neurons are what carry information and spark activity in the brain. Our brain reactions are just neurons firing electrical and chemical signals. On a very basic level, neurons "respond" preferentially to specific stimuli. There is an area called MT in the visual cortex that responds almost exclusively to movement. Within MT, there are neurons that respond to specific types of movement. Some neurons may fire when a stimulus moves from left to right, but not from right to left.
There's a saying in Neuroscience-- "Cells that fire together, wire together." So we have clusters of neurons grouped together that respond to a specific action. Say a subject views a red ball bouncing up and down. The neurons that prefer the color red will fire more strongly than neurons that prefer blue. The neurons that prefer up/down movement will also respond more strongly than ones that prefer right/left. Subjects will give consistent activation of this neurons every time they view that bouncing red ball.
Scientists can record what this activation is. They sum the activity of all the neurons and see which ones are responding more strongly-- this is called population coding. After many trials, they can determine that Cluster A of Neurons likes the color red, and Cluster B of neurons likes up/down motion.
So say in an experiment, researchers play random clips of simple images and record the activity. Using the population coding, scientists can "read" the activation of the neurons and pretty accurately say that a subject is viewing a red ball that is bouncing up and down.
Things are far more fuzzy when it gets complex, however. What about overlapping images? What about pictures of people? What about different sizes and textures of the stimuli? What about emotional reactions to an image?