Showing posts with label causation. Show all posts
Showing posts with label causation. Show all posts

Saturday, March 17, 2012

Weekend Post - You can't trust your eyes. Or your brain.

You can’t trust your eyes. In fact you can’t trust any of your senses. Or even your brain.

One of the most interesting areas of scientific investigation is the study of perception. I remember that as a teenager I became moderately obsessed by visual illusions and wondering how my brain could so easily be fooled.

The image here is a good example. Which of the two horizontal lines is longest? The surprising answer, even when you know it’s true, is that both horizontal lines are exactly the same length. But that’s not what your brain tells you. Your brain demonstrates quite how easily it can be fooled when it fails to correctly interpret this sort of illusion.

There are hundreds of similar examples of such illusions, some much more complicated than this one. They all exploit the sort of perceptual mechanisms that evolved over millions of years that might have been useful in the jungle or savannah where our species originated but aren’t useful any longer in the world of newspapers, smartphones and the Internet.

The great realization is that illusions like the “Müller-Lyer illusion” shown above are just the tip of the iceberg. It’s not just our vision that can be deceived, it’s our entire brain. That’s when an illusion transforms itself into that much more dangerous animal, a delusion.

One of the most common delusions we can all easily experience is called “confirmation bias”. The Skeptic’s Dictionary describes this as “a type of selective thinking whereby one tends to notice and to look for what confirms one's beliefs, and to ignore, not look for, or undervalue the relevance of what contradicts one's beliefs.” More simply, it’s the mistake of noticing and remembering things that confirm our prejudices and ignoring, or simply not seeing the things that contradict them.

I had a conversation with someone not too long ago who suggested that religious believers are more charitable than non-believers. He seemed genuinely surprised when I pointed out that perhaps the two most charitable people in the world, Bill Gates and Warren Buffett both describe themselves as agnostics. He admitted that he knew this but had “conveniently” forgotten it. I think it’s plausible that this was because these facts didn’t fit in with his worldview. Similarly, it’s easy enough for those of us who describe ourselves as atheists to forget that the vast majority of Catholic priests, for instance, are decent people and not child abusers. We remember what we find convenient to remember, not what’s inconvenient. That’s a delusion.

Another example is homeopathy. Every piece of scientific research into homeopathy has confirmed what common sense suggests. It’s nonsense and has no effect. However there are vast numbers of people who swear by it based on anecdotes like “I had a cold, took a homeopathic remedy and within about a week I was better!”

Of course they were better. That’s how colds work, they go away after about a week or 10 days all by themselves whether you take a bogus remedy or not. People forget that but remember the time that they took the useless homeopathic remedy. Believers in so-called complimentary therapies remember the times that the delusional beliefs were falsely “confirmed”.

We also make the common mistake of assuming that correlation implies causation. Because two things happen at the same time or in parallel we assume that they must be connected and one must cause the other. A good example is the observation that increases in medical knowledge in the last century have coincided with a dramatic increase in the number of deaths from cancer. Does this mean that better medicine causes cancer? Of course not, all that’s happening is that the greater lifespan brought about by better medicine has caused people to live longer and older people are much more likely to die of cancer. Cancer is an old person’s disease. Now that we’re much less likely to die from smallpox, typhoid or violence we’re dying of different things instead.

In the same way, the assumption that more and more older people are dying of cancer is somehow caused by the new presence of cellphone masts is another delusion. There is no evidence that cellphone “radiation” causes cancer. In fact the evidence suggests that it doesn’t but it’s extremely human to notice that your aged relative died shortly after the local mast was erected and that one must have caused the other. Tragically your aged relative was going to die anyway.

The good news is that simply knowing about human psychology goes a long way towards combatting these delusions. The better we understand how our brains work (and sometimes fail) the better we can avoid dangerous delusions.

Saturday, November 19, 2011

Weekend Post - What do modern medicine, flu, ice cream and pirates have in common?

Here are some dangerous but factual observations.

Since the introduction of modern healthcare in Botswana more people than ever before have died of cancer. People suffer from more infectious diseases in colder seasons. The more ice cream that is consumed, the higher the number of drownings that occur. The reduction in the number of pirates (yes, I said pirates) in the world has coincided with gradually increasing global temperatures.

All of these are facts, there is overwhelming evidence to support them. But the big question is what they actually mean. Is it true that modern healthcare causes cancer? Does cold weather cause colds and flu? Does ice cream cause people to drown? Has the loss of pirates actually caused global warming? Should any of us actually do anything, should we take any lifestyle decisions based on these undeniable facts?

Of course not, but that’s where the danger lies. You only have to hear of these facts for a chill to go through you. It’s only one step further before you begin to suspect the benefits of modern medicine and science.

The real facts are simple. More people are dying of cancer since the introduction of modern, scientifically based medicine because we’re living longer. Cancer is largely a disease of older people. As we’ve done away with so many of the causes of death from the past that killed younger people like infectious disease and starvation other causes of death have stepped in to fill the void. Like cancer.

People suffer more colds and flu in the winter, not because our immune systems are weaker in the cold months or because viruses are stronger but simply because when we’re cold we stay inside much more. When we’re stuck inside our homes with our families we’re much more likely to spread infections by close-proximity coughs and sneezes.

The really good news is that ice cream doesn’t make you more likely to drown. However, both rates of drowning and consumption of ice cream are both associated with hot weather. Both are caused by something else.

One of the easiest mistakes to make when looking at the world is to assume that correlation implies causation. Because two observations are somehow related then one must have caused the other. It’s a fundamental mistake and one of many logical fallacies that people often make.

A recent example. A couple of weeks ago the BBC reported that scientists had announced in the British Medical Journal “a link” between use of the oral contraceptive pill and prostate cancer. The interesting thing is that only women take the pill and only men get prostate cancer so how can one possibly cause the other? The suggestion is that the hormone oestrogen in the contraceptive pill may be released in the urine of pill-taking women into the water supply. Men then drink the water and experience one of the side-effects of oestrogen: higher rates of cancer. The team of researchers from Canada had carefully examined various countries that used of different forms of contraception and found higher levels of prostate cancer in those countries where the pill was widely used. In countries where other forms of contraception were widespread the higher levels of cancer couldn’t be found.

However they were very careful to say that their study “has significant limitations with respect to causal inference”. In other words they are most certainly NOT saying that the pill increases the risk of prostate cancer. This was just what they call “an ecological study”. They observed a correlation between two things and said that “it must be considered hypothesis generating, and thought provoking”. This wasn’t an experiment, it was just some careful observations that make scientists want to ponder a bit and then design more detailed studies.

One problem is that it’s very difficult to design an experiment to test this. If we suspect there’s a link between oestrogen and prostate cancer we can’t just expose a group of men to oestrogen and see if they develop cancer, that would be inhumane. However it would be worth conducting animal tests and further ecological studies. For instance, does the rate of prostate cancer go down in countries where the use of the pill declines or where lower oestrogen pills are used?

Finally, in case you were worried, nobody really believes there’s a relationship between the number of pirates and global warming. However if you’re willing to suspend your skeptical faculties you can join the Church of the Flying Spaghetti Monster for whom this is a core belief. Perhaps not entirely seriously!

Sources

For various sources on "Correlation Does Not Imply Causation" see the Skeptics Dictionary chapter here (180kb pdf download) or the Wikipedia page here. There's also a comprehensive list of the many logical fallacies courtesy of the Skeptic's Guide to the Universe here.

The story from the BBC about the pill and prostate cancer can be seen here and the original article in the British Medical Journal here.

For the Church of the Flying Spaghetti Monster just say "Arrrgh".