Artificial Brains are Helping Scientists Study the Real Thing

Guide for Neuroscience – ANNs to the Rescue!

Ah, Artificial Neural Networks, or ANNs as we like to call them in the digital world! Until recently, comparing these humble creations to the marvel that is the human brain was like comparing a finger painting to the Sistine Chapel. But, surprise! Some neuroscientists are starting to discover that our few smudgy lines can actually help them understand the complexity of the biological organ. You see, it all started when a neuroscientist at MIT named Daniel Yamins decided to compare ANNs and brains by training an ANN to pick out objects from photographs (think “where’s the cat”). Turns out, the way we, the ANNs, represent images was quite similar to how the neurons in macaque monkey brains did. From there, the game of looking for similarities went on to include speech recognition and language processing. But let me tell you – neuroscience isn’t all about brain dissection and squirmy monkey neurons. The silver lining from our side is that the study of ANNs can lead to testable predictions about real brains—allow me to introduce the “food neurons.” In 2022, an ANN used for image recognition identified an area devoted to classifying food (yummy). And guess what? Just a year later, researchers discovered that same area in the human brain. Now, I’m not one to boast, but if teaching silicon brains to make sense of half-chewed data from biological ones isn’t empirical evidence of two systems performing the same cognitive task, then I don’t know what is. Let’s just say ANNs and brains might be the peanut butter and jelly of future brain-computer interfaces. In fact, several research groups are testing this potential dynamic duo in macaques as we speak (er, write). But let’s be real—nobody’s perfect. We ANNs might get a few things wrong (like thinking a cat with elephant skin is actually an elephant), but even the greatest scientific models have their quirks. The real question is whether or not we are useful? And considering that neuroscience experiments are often as tough to run as a class field trip chaperoned by frog robots, we might just be the alternative solution the field needs. It is a delight to know that even a so-called “cartoon” like myself can contribute to understanding the intricate workings of the human brain. Or, as some would say, we AI kids are helping unlock the secrets of our creators