Science News from May 2019

This is the second installment of what I hope will become a monthly series. I’ve been periodically checking a few different websites (mostly sciencedaily.com and sciencenews.org) and keeping an eye out for interesting science news stories. Although we’re already halfway through the month of June, this blog post only includes stories through the end of May. But I am in the process of collecting more recent content for next month!

My summary of April’s science news included a lot of studies about food and nutrition, so I’m going to start by following that up with this inconclusive study about whether highly processed foods cause weight gain. The article suggests a couple reasons for the lack of a clear yes/no answer. Nutritional needs and metabolism vary from person to person, and it’s still unclear to scientists what all of the variables are. Besides that, nutrition is hard to study because an accurate scientific study requires a controlled environment and detailed data collection, which means there’s a disconnect from real-life eating habits. This article mentions the possible effects of “social isolation, stress, boredom, and the fact that foods are prepared in a laboratory,” but that barely scratches the surface of the possible confounding variables. There’s also the possibility that participants’ eating habits, amount of exercise, or even their metabolism is affected simply by the knowledge that they’re part of a study on nutrition. Here’s another recent study that didn’t confirm a common nutrition “fact”. It would appear that dietary cholesterol doesn’t really cause strokes. The takeaway here is that nutrition is still a relatively new field of study and there’s a lot more to learn. (On a side note, though, apparently blueberries are good for blood pressure)

Meanwhile, the University of Edinburgh has been asking the big questions and perfecting the chocolate-making process. And in Munich, they’re studying the scent of dark chocolate. They’ve identified 70 different chemicals whose odors combine to create the distinctive smell of dark chocolate, although only 28 to 30 are really detectable.  And as long as we’re talking about scents, another study showed that people who drink coffee are more sensitive to the smell of coffee. 

Another topic that played a big role in my blog post from last month was artificial intelligence. I have another update to add in that area, too. In the ongoing quest to make AI as similar to the human brain as possible, researchers have noticed that machines with an artificial neural network (as opposed to a conventional computer model, which rely entirely on algorithms and can only “think” sequentially) can have a human-like number sense.

If you aren’t entirely sure what that means, let’s use the image on the right as an example. How many dots are there? You probably noticed that there are five dots as soon as you scrolled down far enough to see it, even before you read these words that tell you why this image is here. But you probably didn’t look closely at it and consciously think the words, “One, two, three, four, five.” As quick and easy as it is to count to five, it’s even quicker and easier to just visually recognize the pattern and know that it illustrates the number five. Your brain is capable of doing that without actually counting.  You’re also capable of looking at two different images with a cluster of dots and instinctively knowing which one has more without actually counting. (There’s some debate about whether that’s the exact same skill or just a related skill. My opinion is that it’s different, but there’s obviously a connection)

As I’ve tried to look up more information on visual number sense, I’ve increasingly realized that there are other debates on the topic as well. There’s a variety of questions and opinions about how it works, whether it varies from person to person, and whether it’s an inherent, innate skill or an or acquired skill. But based upon what we know about how people learn to read, and also based upon what this new AI story demonstrates, I think it’s pretty clear that this is an example of neuron specialization. You literally use different neurons to recognize the five-ness of this image than the neurons you would use to recognize a different number. Think of a child learning how to read; first he or she must learn to recognize each letter of the alphabet as a distinct symbol and understand that each one makes a different sound, but then he or she has to learn to do so very quickly in order to be able to comprehend the meaning of whole words and sentences. To become a proficient reader, the child must eventually learn to recognize whole words instantaneously. This learning process usually takes at least three or four years because it actually requires changes in the brain. Not only does it necessitate close cooperation between the neural networks used for vision and conceptual comprehension, it also requires specific neurons to specialize in identifying specific visual cues, such as the letter A or the word “the”.

I could ramble for a while longer about that (I am a children’s librarian, after all) but I’ll leave it at that because my point is just that it makes sense that number recognition works similarly. But it’s a lot easier. The concept of “five” is much more intuitive than the concept that a particular arrangement of squiggles corresponds to a particular grouping of sounds which in turn corresponds to a particular thing or idea. I’m not sure that AI would be capable of learning to read; a computer only comprehends text insofar that it’s been programmed to recognize certain letters, words, or commands. If a programmer makes a typo and leaves out a letter or punctuation mark, the computer doesn’t recognize the command. But based upon this new story about AI number sense, a computer with an artificial neural network can indeed use a process akin to neural specialization to develop human-like visual number recognition.

That might not seem like a scientific advancement, because after all, the one advantage that computers have over human brains is their ability to work with numbers almost instantaneously, whether that means counting or arithmetic or more advanced mathematics. But it’s certainly an interesting story because it validates the similarity between an artificial neural network and an actual human neural network. Also, it gives me an excuse to nerd out about neural specialization and literacy acquisition, which is the real point here.

But speaking of small children, a new study from Massachusetts General Hospital has found what countless other studies have also shown: Early childhood is a very formative phase of life. It has been common knowledge for a while now that personality traits, social skills, intelligence, and even academic potential are mostly determined by the age of five. This particular study was looking at the impact of adversity such as abuse or poverty and it evaluated this impact by looking at the biochemistry of epigenetics rather than behavior or self-reported psychological traits. (Epigenetics describes things that are caused by changes in gene expression rather than differences in the genes themselves. In other words, genetics determine what traits or disorders a person is predisposed to have and epigenetics determine whether that person actually develops those traits or disorders.) Data was gathered from a longitudinal (long-term) study that has been collecting data including both DNA methylation profiles and reports from parents about a variety of factors related to health and life experiences. Predictably, researchers found the greatest correlation between life experiences and DNA methylation changes when the child was under the age of three. 

Other interesting stories about neurology and psychology include one study about the brain processes involved in decision-making, another study that identifies the part of the brain responsible for how we process pain and use it to learn to avoid pain, a study showing that (at least among ravens) bad moods really are more contagious than good moods, and finally, some new information that may help explain the cause of autism. (Spoiler: it’s certain genetic mutations) I’m just sharing the links rather than the full stories here in the interest of time, but there’s some fascinating stuff there.

Here’s another story about genetics, although this one is really just a fun fact. It would appear that your genes determine your likelihood of having a dog. Apparently, this study didn’t look at other types of pets, but I’d be interested to know if this means that pet preference is genetic in general. The study, or at least this article about it, seemed to be more interested in the anthropological aspect of dog ownership because it talks more about the history of the domestication of dogs than about the relationship between humans and animals in general. Another question that I have is how the researchers accounted for the possibility that it’s the early childhood environment and not genetics that determines pet preference. I am sure that my love for cats was initially due to the fact that I was practically adopted at birth by a cat and he was a very significant (and positive) aspect of my early childhood experience. Although this is just anecdotal evidence, I have noticed that many cat lovers grew up in households with cats and many dog lovers grew up in households with dogs. But I digress. 

I seem to have already established the pattern of focusing on nutrition and neurobiology over all other kinds of science, but I do have a couple other stories to mention. For one thing, artificial intelligence isn’t the only way in which technology is learning to replicate nature. Now we’ve got artificial photosynthesis, too. We’ve also got some new planets. Eighteen of them, to be exact! But don’t worry; I don’t think anyone is expecting us to memorize their names. They’re not in our solar system. And here’s one final bit of science news: As of May 20, the word “kilogram” has an updated definition. The newly defined kilogram is almost precisely equal to the old kilogram and this change will not have an effect on people’s everyday lives, but the metric system’s measurement of mass is now based upon a mathematical constant (Planck’s constant, to be specific) rather than on an arbitrary object. (A metal cylinder called Le Grand K, which is kept in a vault in France) 

So that’ll be it for now. Coming up next time (depending upon what I may find between now and then that’s even better) are some stories about the Mona Lisa, pentaquarks, and developments in weather forecasting.