Something has puzzled me for a long time about the psychology of those who deny climate change—about the denialists, as they’re called. I’m talking about the serious climate change deniers, the ones who go around making “research” presentations on the matter, like Lord Monckton. But I think I’ve just recently started to grasp what’s going on in their heads.

Denialists seem somewhat rational at first: they form denials in accordance with the fragment of evidence that they actually have seen, even if the overall weight of the evidence is against them. On this view, there’s not much of a puzzle. The denial is rational in light of the evidence that they know. Perhaps they think thoughts like this: ‘Richard Lindzen holds that climate change hasn’t been proven, so—given his scientific authority—it is rational for me not to believe in it.’ Or: ‘The polar ice caps grew in size from September 2007 to September 2009, so it is rational to believe they are not melting.’ Is this what’s going on? Let’s call this the Rationalizing Reconstruction of denialist psychology: the fragment of evidence they know causes the denier’s belief.

If the Rationalizing Reconstruction is right, denialists simply need to be cured by more evidence. They only have some evidence—so the story goes—but a broader array will fix this and hence fix their apparently irrational beliefs. Point out to them that the overall size of the ice caps has decreased 20% since 1979. Point out that there are many more scientists who affirm that climate change has been demonstrated to a high degree of confidence. And in light of this evidence, they’ll change their minds, right?

The problem with this model is that we know its hoped for solution doesn’t work. As Dan Kahan has shown, simply giving denialists more evidence doesn’t get them to change their minds. And perhaps we should have expected this, in light of the results on motivated reasoning that Ziva Kunda brought to light in 1990.

The natural thing to say, of course, is that the evidence just doesn’t matter to the denialists. They don’t care about it one bit. Call this the Pure Motivation view of denialist psychology.

But this view leaves something out. Many climate change deniers have an intense focus on various bits evidence about climate change. Lord Monckton and James Inhofe, in their presentations, do come equipped with pertinent facts, albeit facts that have been cherry picked and misleadingly arranged. And the Pure Motivation is unable to account for this fascination with at least some of the evidence.

So we have a dilemma. If the Rationalizing Reconstruction is correct, we can’t explain why the denialists’ views don’t change in light of more data. If the Pure Motivation view is correct, we can’t explain why they pay attention to evidence as much as they do.

Our dilemma becomes even more pressing when we consider the following sobering fact. In order to have obtained and arranged the evidence that seems to support their cases at all, climate change deniers will have to have sifted through mountains of evidence that stand contrary to their denial.

Think about this for a moment.

In order for Lord Monckton (say) to find the years in which the ice caps grew back some, he would have had to sift through the data of many more years in which they shrank, otherwise—without sifting through that data—he would not have known to throw out the years in which they shrunk. Likewise, in order to find the year out of the last fourteen that is not among the fourteen hottest years on record (2008 at #16), one has to look at the average earth temperatures of the other years, all of which are (2001-2014, minus 2008). In short, only by scouring data that supports climate change can one find the bits of evidence that seem to go against it.

So we should have known from the start that the Rationalizing Reconstruction couldn’t be sustained. The very gathering of the deniers’ “evidence” requires cognition of even more evidence to the contrary of their “beliefs.”

I put “evidence” and “beliefs” in scare quotes in the last paragraph, because I wish to suggest that something else altogether is going on.

Kendall Walton has discussed the role of props in games of make-believe and in fiction more generally. A prop toy horse, for example, is something that prescribes that you imagine a toy horse, according to certain principles of generation in certain games of make-believe, and then interact with it accordingly. You can also assign arbitrary objects prop statuses. To take a well-known example from Walton, you can play a game in which tree-stumps ‘count as’ bears, so that every time you see a tree stump in the woods you must imagine it’s a bear and pretend accordingly.

My suggestion is that the cherry-picked climate data function in the minds of climate denialists in ways similar to props in make-believe. The cherry-picked data constitute a set of props for the denialist. The principle of generation then is for the denialist to form ideological imaginings that comport with those props. This gives the denialists their semblance of rationality. They draw trend lines and make inferences in accordance with the props they have chosen, just as the children in the woods draw inferences about the concentration of ‘bears’ in various parts of the woods.

How willful is this process? And how good is the denialists’ metacognition of the game of make-believe that is going on in their own minds? I’m not sure we know the answers to these questions. The murky answers for now must be “semi-willful” to the first question and “poor” to the second.

But I think we can say that the game they are playing is motivated—typically—by the need to belong to a certain in-group (in this case: conservatives); that’s why Kahan uses the phrase “identity-protective cognition.” Thus to defeat the denialists, we can’t just lay out more evidence—the new pieces of data simply won’t be taken up as props—we have to motivate them to play a different game. How to do that, as far as I can tell, is completely unknown. But at least the morally important task is a bit clearer.