Random

There's a good post over at Bad Science about how bad people are at seeing real patterns.

People are great at seeing patterns, it's one of the things we do really well. But sometimes were too good, and we see patterns that aren't there. For example, we see faces everywhere we look. Just Google image search "ghosts" and you'll see a whole bunch of shadows that are interpreted as faces (as well as photoshopped images). Another good example of this is the well-known face on Mars, which is just a plateau that was photographed at a convenient time, the face was in the shadows, not the planet.

An example given in the Bad Science piece is "runs", and why random patterns can look decidedly nonrandom. He gives a specific experiment that demonstrates this effect, but there's another well-known one. If you've taken a class in or read any material on probability theory, you've probably heard this story. It goes that a Professor of a large introductory probability class gives an odd assignment the first day of class: half the class flips a coin 100 times and records the result, while the other half just makes up data. He goes through the results and can separate most of the people who made up data from those who didn't.

How would this be done? As it turns out, people hate putting down long strings of the same result, but in a large data set they're virtually guaranteed to occur. So the results that contain strings of 6 or so in a row are usually real, and those that don't aren't. Of course, this is probabilistic, it doesn't always work, but it does with pretty good probability.

Not only are people horrible at creating random patterns, we're just as bad at making random numbers. Some folks over at Scienceblogs (specifically Cognitive Daily) did an experiment to demonstrate this, and it has been verified by others in the science blogging community.

The result is that if you ask people, 17 is the most random number (between 1 and 20 anyway). I guess it "feels" more random, whatever that means.

I think this can be generalized even further, in that we really like for there to be patterns behind things, for events to have causes that aren't random. Einstein famously said, "God doesn't play dice" in voicing his opposition to Quantum Theory, but indeed, the universe is probabilistic, whether we like it or not (according to a new study, nature might be even weirder).

This is why, I think, we have such a hard time accepting that certain events were random. People still think that the Kennedy Assassination was a conspiracy theory, despite thorough debunking of all conspirator's claims (see, for example, Penn & Teller's episode of Bullshit! on it). I'd even go so far as to say it's behind the reactions to the VT massacre earlier this week (at least those not obviously motivated by grinding axes), and the seemingly universal desire to believe that there's a greater cause driving the world, that things happen for a reason.

We see patterns everywhere, and so they must be everywhere. It's too bad that sometimes the world just is random, and we have to learn to deal with that.

Antibiotic interactions that select against resistance

There was a great paper in Nature this week, called Antibiotic interactions that select against resistance by Remy Chait, Allison Craney and Roy Kishony (subscription required). I’m frankly completely surprised that it’s not plastered all over the place, given how cool it is.

Here’s the abstract (citations removed):

Multidrug combinations are increasingly important in combating the spread of antibiotic-resistance in bacterial pathogens. On a broader scale, such combinations are also important in understanding microbial ecology and evolution. Although the effects of multidrug combinations on bacterial growth have been studied extensively, relatively little is known about their impact on the differential selection between sensitive and resistant bacterial populations. Normally, the presence of a drug confers an advantage on its resistant mutants in competition with the sensitive wild-type population. Here we show, by using a direct competition assay between doxycycline-resistant and doxycycline-sensitive Escherichia coli, that this differential selection can be inverted in a hyper-antagonistic [that is suppressive, the drugs are less effective together than individually] class of drug combinations. Used in such a combination, a drug can render the combined treatment selective against the drug's own resistance allele. Further, this inversion of selection seems largely insensitive to the underlying resistance mechanism and occurs, at sublethal concentrations, while maintaining inhibition of the wild type. These seemingly paradoxical results can be rationalized in terms of a simple geometric argument. Our findings demonstrate a previously unappreciated feature of the fitness landscape for the evolution of resistance and point to a trade-off between the effect of drug interactions on absolute potency and the relative competitive selection that they impose on emerging resistant populations.

That’s quite dense, and I’ll readily admit that I don’t understand everything they say in the paper itself, but the general gist is that there are combinations of drugs that will actually select against a resistant mutant rather than the normal (called ‘wild’) type. This doesn’t seem to make any sense, but it works because different concentrations of the two drugs promote the growth of the resistant and wild types differently, producing a spectrum containing spots where the wild type “wins” despite being more sensitive to the drugs on their own.

The way this works is quite complicated (and where I get lost), but it has to do with the fact that when drugs interact they don’t just do the sum of what they would normally do. They can synergize and become more effective, or suppress each other and become vastly less effective. The latter is responsible for preferential selection against the resistant type. In the authors’ words, “Our data show that in suppressing drug combinations, a drug can be used to exert competitive selection against its own resistance allele.” The biological rational for this is quite complicated and technical, filled with impenetrable jargon. However, none of that diminishes the striking result suppressive drug combinations can select against bacteria when they would select for those bacteria individually.

The authors insist that these findings are preliminary and that more research needs to be done. They only looked at one bacterium’s resistance to one drug, and attempting to generalize these results any further could be very misleading. However, this could open a new way to treat drug-resistant bacteria, which is a huge problem for modern medicine. This is a very cool experiment and result, and one of those papers that made me go, “hooray science!”