Doing Good Science

Doing Good Science

Building knowledge, training new scientists, sharing a world.

Brief thoughts on uncertainty.


For context, these thoughts follow upon a very good session at ScienceOnline Together 2014 on "How to communicate uncertainty with the brevity that online communication requires." Two of the participants in the session used Storify to collect tweets of the discussion (here and here).

About a month later, this does less to answer the question of the session title than to give you a peek into my thoughts about science and uncertainty. This may be what you've come to expect of me.

Humans are uncomfortable with uncertainty, at least in those moments when we notice it and where we have to make decisions that have more than entertainment value riding on them. We'd rather have certainty, since that makes it easier to enact plans that won't be thwarted.

Science is (probably) a response to our desire for more certainty. Finding natural explanations for natural phenomena, stable patterns in our experience, gives us a handle on our world and what we can expect from it that's less capricious than "the gods are in a mood today."

But the scientific method isn't magic. It's a tool that cranks out explanations of what's happened, predictions of what's coming up, based on observations made by humans with our fallible human senses.

The fallibility of those human senses (plus things like the trickiness of being certain you're awake and not dreaming) was (probably) what drove philosopher René Descartes in his famous Meditations, the work that yielded the conclusion "I think, therefore I am" and that featured not one but two proofs of the existence of a God who is not a deceiver. Descartes was not pursuing a theological project here. Rather, he was trying to explain how empirical science -- science relying on all kinds of observations made by fallible humans with their fallible senses -- could possibly build reliable knowledge. Trying to put empirical science on firm foundations, he engaged in his "method of doubt" to locate some solid place to stand, some thing that could not be doubted. That something was "I think, therefore I am" -- in other words, if I'm here doubting that my experience is reliable, that I'm awake instead of dreaming, that I'm a human being rather than a brain in a vat, I can at least me sure that there exists a thinking thing that's doing the doubting.

From this fact that could not be doubted, Descartes tried to climb back out of that pit of doubt and to work out the extent to which we could trust our senses (and the ways in which our sense were likely to mislead us). This involved those two proofs of the existence of a God who is not a deceiver, plus a whole complicated story of minds and brain communicating with each other (via the wiggling of our pineal glands) -- which is to say, it was not entirely persuasive. Still, it was all in the service of getting us more certainty from our empirical science.

Certainty and its limits are at the heart of another piece of philosophy, "the problem of induction," this one most closely associated with David Hume. The problem here rests on our basic inability to be certain that what we have so far observed of our world will be a reliable guide to what we haven't observed yet, that the future will be like the past. Observing a hundred, or a thousand, or a million ravens that are black is not enough for us to conclude with absolute certainty that the ravens we haven't yet observed must also be black. Just because the sun rose today, and yesterday, and everyday through recorded human history to date does not guarantee that it will rise tomorrow.

But while Hume pointed out the limits of what we could conclude with certainty from our observations at any given moment -- limits which impelled Karl Popper to assert that the scientific attitude was one of trying to prove hypotheses false rather than seeking support for them -- he also acknowledged our almost irresistible inclination to believe that the future will be like the past, that the patterns of our experience so far will be repeated in the parts of the world still waiting for us to experience them. Logic can't guarantee these patterns will persist, but our expectations (especially in cases where we have oodles of very consistent observations) feel like certainty.

Scientists are trained to recognize the limits of their certainty when they draw conclusions, offer explanations, make predictions. They are officially on the hook to acknowledge their knowledge claims as tentative, likely to be updated in the light of further information.

This care in acknowledging the limits of what careful observation and logical inference guarantee us can make it appear to people who don't obsess over uncertainties in everyday life that scientists don't know what's going on. But the existence of some amount of uncertainty does not mean we have no idea what's going on, no clue what's likely to happen next.

What non-scientists who dismiss scientific knowledge claims on the basis of acknowledged uncertainty forget is that making decisions in the face of uncertainty is the human condition. We do it all the time. If we didn't, we'd make no decisions at all (or else we'd be living a sustained lie about how clearly we see into our future).

Strangely, though, we seem to have a hard time reconciling our everyday pragmatism about everyday uncertainty with our suspicion about the uncertainties scientists flag in the knowledge they share with us. Maybe we're making the jump from viewing scientific knowledge as reliable to demanding that it be perfect. Or maybe we're just not very reflective about how easily we navigate uncertainty in our everyday decision-making.

I see this firsthand when my "Ethics in Science" students grapple with ethics case studies. At first they are freaked out by the missing details, the less-than-perfect information about what will happen if the protagonist does X or if she does Y instead. How can we make good decisions about what the protagonist should do if we can't be certain about those potential outcomes?

My answer to them: The same way we do in real life, whose future we can't see with any more certainty.

When there's more riding on our decisions, we're more likely to notice the gaps in the information that informs those decisions, the uncertainty inherent in the outcomes that will follow on what we decide. But we never have perfect information, and neither do scientists. That doesn't mean our decision-making is hopeless, just that we need to get comfortable making do with the certainty we have.

The views expressed are those of the author and are not necessarily those of Scientific American.

Share this Article:


You must sign in or register as a member to submit a comment.

Starting Thanksgiving

Enter code: HOLIDAY 2015
at checkout

Get 20% off now! >


Email this Article