In this study, we continuously monitored Egyptian fruit bats for months, recording audio and video around-the-clock. We analyzed almost 15,000 vocalizations, which accompanied the everyday interactions of the bats, and were all directed toward specific individuals, rather than broadcast. We found that bat vocalizations carry ample information about the identity of the emitter, the context of the call, the behavioral response to the call, and even the call’s addressee. Our results underline the importance of studying the mundane, pairwise, directed, vocal interactions of animals.

This is brilliant. They were able to correlate their data analysis of the bats’ vocalizations with the behavior and responses that they observed… so now we know more about how bats communicate! Simply by listening to the vocalization, the context, addressee, and even “the outcome of the interaction can be predicted above chance level”. Fascinating.

From the discussion:

It is important to note that we used one set of acoustic features for classification. However, many other multi-dimensional spectro-temporal representations can be tested. The bat’s brain could thus be using some other representation that encapsulates much more information regarding different social aspects. The bat may be able to classify the context of an interaction with higher confidence, based on some acoustic feature which it evolved to use and is yet to be determined. Our analysis is thus probably only a lower bound on what a bat is capable of extracting from aggressive social vocalizations. For example, we did not include any temporal information in our analysis.

In any acoustic signal, and especially where communication is involved, the time parameter is usually crucial and will add rich layers of information. For example, just imagine taking a piece of human speech, and (a) only looking at the overal speech parameters, versus (b) observing how the speech parameters change during the speech. Case (b) will provide far more information than case (a). I think we will discover over time that bats have a pretty well-evolved communication scheme.

This is fascinating stuff.

Bayesian inference is a way to get sharper predictions from your data. It’s particularly useful when you don’t have as much data as you would like and want to juice every last bit of predictive strength from it.

Although it is sometimes described with reverence, Bayesian inference isn’t magic or mystical. And even though the math under the hood can get dense, the concepts behind it are completely accessible. In brief, Bayesian inference lets you draw stronger conclusions from your data by folding in what you already know about the answer.

An excellent, simple introduction to Bayesian inference. This uses practical examples and an abundance of visual guides: especially useful if you don’t have an extensive math background.

One of my pet peeves with scientific journalism is the propensity to confuse correlation with causation. The idea is that just because two things are observed to happen at the same time (or before, or after, one another), does not imply that one causes the other.

In the latest example of this, the link between chocolate and good health is revisited.

The article opens with:

People who eat chocolate regularly tend to be thinner, new research suggests.

… which implies that a causation has been observed. The article goes on to make the following points:

[…] those who ate chocolate a few times a week were, on average, slimmer than those who ate it occasionally.

The link remained even when other factors, like how much exercise individuals did, were taken into account.

[…] it is how often you eat chocolate that is important, rather than how much of it you eat. The study found no link with quantity consumed.

So… I’d still lose weight if I ate a tonne of chocolate very frequently? Really?!

The most important statement, however, comes a little later:

But the findings only suggest a link - not proof that one factor causes the other.

… and,

And if you are looking to change your diet, you are likely to benefit most from eating more fresh fruits and vegetables.

Now guess what the headline of this article, which itself says that it’s only a link, and talks about maintaining an overall good diet, reads.

Chocolate ‘may help keep people slim’

Perfect, isn’t it?

As is well known, Earth plays host to numerous meteors, some of which are big enough to reach the Earth’s surface as meteorites. What scares us, the human species as a whole, of course, are the ones large enough to cause significant damage—especially the ones that can cause mass extinctions.

Are the rates at which meteorites arrive at Earth cyclic? Can we predict when the next mass extinction (meaning almost certainly the end of the human race as we know it) will be upon us? It’s all in the statistics!

Of course, there’s no other way make predictions, other than to watch for patterns and extrapolate the timescales involved into the future. Based on past records, there are certain hypotheses about when and why some big rocks hurl towards the Earth—and the most interesting of them involves a phantom solar companion. Christened “Nemesis”, this hypothetical star is supposed to be lurking at the outer edges of the solar system, and like a classical villain hurling chunks of rock from the [Oort Cloud][linkoort] into the inner solar system. Essentially, he’s made out to be playing Duck Hunt, with Earth as the Duck, of course.

But all of this story (and other cyclical hypotheses) depends on a robust analysis of data, which involves robust statistical methods. Everything is a question of probability, of course—you can’t get absolute numbers in situations like these—and it’s no easy matter to set limits to what the error tolerances are. To understand how difficult the whole process is, consider the data in this case: cataloguing previous impact craters on Earth against how old they are. There are numerous craters, of course, but there are also erosion processes that make life difficult.

Well, according to this article by Ian O’Neill, there isn’t much statistical evidence after all for Mr. Nemesis. Researchers at the Max Plank Institute of Astronomy turned to Bayesian probabilities to look at the data, and concluded that there isn’t periodicity in the data as previously thought. Bayesian probabilities are a little more complex implementation of probability theory, where the question asked is: “What is the probability of event A, given that event B precedes it?” Event B, of course, has its own probability of occurring. (There are other complexities, of course, but that’s the simplest case.)

We’re not doomed by Nemesis after all! Apparently robust statistics has saved the day once again. :)

Or has it? The scientists did not find cyclical evidence that would point to Nemesis, but they think they’ve found evidence for an increasing rate of meteorites over the past 250 million years. And before you say “erosion!”–which would decrease the number of older craters—apparently the same trend is observed on the Moon, where there is, of course, no erosion going on.

So—we’re not out the woods yet! ;)

(In case you’re seriously wondering—no, there isn’t a serious chance that we’d be hit by a big one soon. There isn’t even a less serious chance. Doesn’t mean it’s impossible, though!)