Hume and the problem of induction

In 2014, Japanese scientist Haruko Obokata announced a startling discovery in one of the world’s most prestigious scientific journals, Nature. She had found that she could create pluripotent stem cells out of ordinary cells using certain types of stress, like acid or even physical trauma. Pluripotent stem cells have the ability to become any type of cell, and scientists hope that, in the future, injuries of any sort will be healed with the simple application of some of these stem cells.

Although we had already known how to create pluripotent stem cells before Obakata’s discovery, it was a difficult, expensive, and somewhat ethically questionable process, involving the use of embryos. Obokata’s method was easy, cheap, and completely ethical. All you’d need was some cells and some acid. If it proved replicable, doctors of the future would be able to provide their patients with stem cells on demand, and medicine as we know it would be forever changed. You might even say it would undergo a paradigm shift.

You might be wondering at this point why Haruko Obokata isn’t a household name. Or, if you’re more familiar with science news, you might be shaking your head already. The short answer is because she faked her results. She falsified data and then tried to cover her tracks. Her story, so briefly a triumph, ended up a disgrace. Her PhD was revoked, one of her advisors committed suicide, and the organization which provided her with resources, RIKEN, was roundly criticized for allowing such a scientific travesty to take place.

This isn’t a morality tale, however. I’m not concerned with Haruko Obokata, or her advisors. They knew what they were doing, and their failings were ethical ones. I’m concerned with Nature and its editors. The editors of Nature had no idea that Obokata faked data. All they knew was what they were presented with: evidence seeming to show that stress caused ordinary cells to become pluripotent stem cells. Should they have known better?

As you might imagine, there’s a philosopher who had a lot to say about this issue. His name was David Hume, and he was one of the foremost members of the Scottish Enlightenment, an 18th century movement which included among its members Adam Smith and Robbie Burns. Hume was concerned with many things, but what he’s perhaps most famous for is his analysis of the issue of causation.

Aristotle believed that he could prove something caused something else by setting up an elaborate logical system. For example, in his logical system, rocks fell to Earth because they naturally belonged on Earth. Their “belongingness” caused their falling. Bacon disagreed, and insisted that we prove causation by careful observation of the real world. Bacon would hope to discover why rocks fell by observing enough instances of rocks falling, as well as instances of rocks not falling, and then carefully coming up with a theory about when rocks would fall and not fall.

Hume took a different approach. First, he disagreed with Aristotle entirely. In fact, he claimed that there was no such thing as a logical system independent of the real world, because all of our ideas come from our interactions with the real world in some way or another. This is a pretty major claim in philosophy, but we’ll leave it for now. More importantly for our purposes, Hume said that the only way you can hope to approach if something causes something else is by observation, but you will never actually know if something causes something else. In other words, no matter how many rocks Bacon watched, Bacon would never truly know what caused a rock to fall.

This is a serious claim. Someone like Bacon, had he been alive at the same time as Hume, probably would have been furious. But let’s look at Hume’s evidence, first. Hume tells us that what we consider causation is really just two events happening at the same time, or one after the other. So, for instance, if you take a sip of hot tea and burn your tongue, you think, “The hot tea burned my tongue!” Hume says, “No, you took a sip of hot tea and you burnt your tongue. I don’t see why one has to have caused the other.”

The obvious response to this is that every time you’ve taken a sip of burning hot tea in the past, you’ve burnt your tongue. And so you think that it makes logical sense to say that hot tea burns your tongue. But Hume still disagrees. He sees no reason why you should be allowed to assume anything from the past. Hume wants a reason why hot tea necessarily burns your tongue, not that you’ve seen that connection in the past.

Well, the final answer has to be that you’ve always made connections based on what’s happened in the past, right? That’s how you learned everything, and it’s worked so far. You learned how hot tea burns your tongue by doing it as a little kid. You learned how to use pencils by playing with them when you were little, making doodles on a paper. Almost all of your knowledge is composed of these associations you’ve made, that you do something and something else happened.

In other words, according to Hume, in the end your justification for reasoning from the past is that, in the past, reasoning from the past has worked. You think you can predict that the next time you sip hot tea, you will burn your tongue, because this is how you learned your lesson about putting a marshmallow too close to the fire when you were little. In philosophical terms, your only proof of induction is induction.

The implications of this are quite serious. In Cartesian terms, science is built on a shaky foundation, and our impressions of causation are really just associations. Hume even tells us that before we argue with him that our knowledge is different, we have to consider how sure others have been of wrong causations, like tribes who believed their dances caused the rain. Despite all of our advances, there’s no way of proving we’re not just as mistaken as those tribes, about any of our scientific ideas.

Our answer to this problem, according to Hume, is to treat all philosophy as entertainment and to stop teaching it in schools. It’s hard to tell if he’s serious, but it certainly seems somewhat appealing. Unfortunately, we cannot take that option. We have to proceed onwards, always aware of how shaky our foundations are, putting one foot carefully in front of the other on the rickety, occasionally rotten bridge of induction.

Now, back to the editors of Nature, and what they should have done. In my opinion, they should have recognized that, by the standards of induction, Obokata’s claim was extraordinary. Putting cells through stress, whether physical trauma or acid, had to have been something other scientists had done before, and none of them had noticed any pluripotent stem cells. The pattern of association, in other words, was not there. Furthermore, her graphs and figures were unusual for the data, raising further red flags about how carefully she analyzed her data, and how open she was to the possibility that her extraordinary claim was incorrect.

Induction is never totally certain in our world. It merely exists on a scale of certainty, as Hume proved. What is certain is that extraordinary claims require extraordinary evidence to be even tentatively believed, and Obokata made an extraordinary claim with poor evidence. Her claims were so extraordinary that they were in fact more tempting to believe than an ordinary claim, which is a phenomenon that Hume was well aware of. Nevertheless, the editors of Nature should have been on their guard, and realized that Obokata, even if she hadn’t faked her data, should never have claimed that stress caused cells to revert to pluripotent stem cells. Alas, they didn’t, and the scientific world was worse off for it.

There’s a follow-up question that Hume never asked, though. You might wonder why, if causation is never able to be proven, that humans believe it in the first place. In other words, what causes the belief of causation? Hume would say that asking the question proves that we didn’t really understand his critique, but I’d beg to differ. And so would Immanuel Kant, who we’ll cover next time.

Leave a Reply

Your email address will not be published. Required fields are marked *