Info 202 F12

This is an interesting article from Nature about scanning your brain while you are sleeping, in order to decode the visual content of your dreams by classifying the reported content

A team of researchers in Japan conducted functional neuroimaging to scan the brains when the participants were sleeping, and recorded the brain activity by using EEG. They woke the participants up when they detected the brain activity was associated with dreaming, and asked them to talk about their dreams, and then asked them go back to sleep. This procedure was done in 3-hour blocks, repeated 7~10 times on different days. During each block, the participants were woken up 10 times per hour. On average, each participant reported 6~7 visual dreams in each hour, which gave out about 200 samples in total.

The researchers extracted keywords from their verbal report and picked 20 categories with the higher frequencies, and assigned a photo representing each category. Later, the researchers scanned the participants' brain activity while asking them to viewed the photos of the categories. By categorizing and analyzing the brain wave, they found that the content presented in one's dream has strong relation with the visual perception. Furthermore, by classifying the brain activity while one is sleeping, they can predict the content of the dreams. (Mind-reading!)

What is being organized?

The resources being organized here are the different brain wave activity images, provided by the participants.

How is it organized?

Each resource is related with a category that specified by the researchers. The categories were originally given from the participants, however, the researchers chose the top 20 most  appeared words as categories, that might caused biased and insufficient if there were more additional data coming in. Just as in our assignment, the context affected the scope of the vocabularies that we used when assuming there were only 15 instances in the world.

Why is it organized?

There are two main purposes of this study. One is to find out the association between one's brain activity when really looking at something and dreaming about something. The other is try to use this model to predict the content of one's dream.

The study contains two parts of organizing system: 

1) categorizing the existing resources (20 terms/categories), and

2) classifying the resources (brain activity images) into existing categories, in order to predict the content of one's dream.

However I found it interesting that in this study, there were only 3 participants, so the prediction might only falls into specific categories of people. On the other hand, the study didn't explain the variance between the categories, such as whether "car" and "computer" were exclusive, or the approach that combining them together as "utilities" would produce a better result.