What's Truth got to do with it?
Optimality Models, Explanation, and Idealization
21 September 2012
Truth explains and more of the truth explains better. This has the feeling of a secure truism. It lies behind the success of Gilbert Harman's well-chosen slogan of 1965, "The Inference to the Best Explanation." If more truth explains better, then the best explanation has the most truth. Of course we should infer to it.
Collin Rice is visiting as a Postdoctoral Fellow in the Center this year. The goal of his talk today is to weaken this connection of truth and explanation. Sometimes, he is going to tell us, less truth explains better. We might even get the best explanation from a mere grain of truth.
It is noon and I am standing with Collin at the podium, watching his audience file in. There's a cluster around the toaster and bagel tray. It will be a good "house," to use the theatrical jargon. Two of our colleagues from CMU have taken the 15 minute walk over Panther Hollow to join us. When they arrive, I need to say something to Collin.
"Do you know Clark Glymour?" I ask.
"Well you will. He's over there." Somehow it seemed that I hadn't said enough. So I added, "Get ready. He likes to ask colorful questions."
Once the room was called to order, Collin began to speak. His starting point was a reminder of the role truth plays in standard accounts of explanation. He had the faces of two Pittsburgh heroes of explanation on the screen: Carl ("Peter") Hempel and Wes Salmon. They favored truth in a good explanation. There would be more Pittsburgh heroes mentioned: Jim Woodward and Bob Batterman--both of whom happened to be sitting in the room.
The device that would enable Collin to separate explanation and truth is what he has called "optimality models." The details are complicated. But the basic idea is strong and simple.
Optimality models explain by showing that a certain state happens to be optimal. For example, foraging lapwing birds scan for food, move a short distance, scan again, and so on. Why do they move just that distance? In the simplified models of their wandering, that is the optimal distance. Anything less does not do well in exposing new territory for scanning. Anything more ends up being too costly for the gain in terms of the energy spent.
This was one of many examples Collin sketched. They all depended on examining a stripped down, highly idealized model. That fact was key. We learned from their success that all the suppressed details were unimportant for explaining the processes at issue. At the same time, exactly because the model was so stripped down, we had found an explanation of some generality.
Here was the separation of truth and explanation. The more truths we stripped out the better the explanation.
There was a particular sort of truth that was being stripped out. Collin's optimality explanations looked at the average equilibrium behavior. They neglected to give an explicit account of how the systems get to this equilibrium.
There is a long causal story that could be told. It might include the sad fate of the imprudent lapwings that move unnecessarily far between foraging. These careless birds grow leaner from working more for fewer calories and weary from all the effort. When they pause to rest, perhaps lingering a little too long over the latest issue of Bird News, a hungry predator swoops down. Their strategy is extinguished.
© Hans Hillewaert / CC-BY-SA-3.0
The optimality models, Collin assured us, have stripped out these causal stories. They do not figure in his explanations. This claim will draw some attention after the floor was opened to questions.
The talk concluded. In the break, Mark Wilson took a moment with Collin to excuse himself. He was under the pressure of an impending faculty meeting and had to leave. He then took many moments more to explain to Collin some important relation he had discerned between Collin's interests and Mark's latest project. I waited for those moments to pass. I could see that the time for open questions was nearing. I was about to urge a quick close, when Mark signaled the end by asking Collin to email him. I could avoid the awkwardness of asking Mark to hurry up.
Clark Glymour likes a show and knows how to put one on. So I was not at all surprised that he wanted to ask the first question. But then Collin is a freshly-minted PhD and Clark can be a rhetorical steamroller. So I wanted to lighten the weight that would soon roll over Collin.
"You'll keep it short?" I said for lack of anything more effective.
"Well..." Clark chortled, "I have three tiers in my question."
He proceeded through his three tiers with some vigor. First, Collin's examples were all about aggregations of individuals, not individuals. Second, there is causal talk embedded in the models. Natural selection in biology is all about the causal relation of individuals engendering children. Then finally, Clark praised Collin's treatment of the optimality calculations, but wanted to know just what was added by connecting them with explanation.
Collin may be young, but he is quick and articulate. With a smile, he tackled Clark's three tiers, point by point. I found myself following the individual claims. But I was feeling lost on the overall drift of things. As the words bounced to and fro between Clark and Collin, that sense strengthened.
It's not uncommon that I find myself struggling to follow what appears to be a perfectly well-functioning exchange. I'm not so clever or quick. There was something about this one that made me wonder if I was the only one feeling lost. It is a gamble. A chair should intercede only rarely. This seemed like a good time; and it might help a junior scholar in his tussle with a senior one.
"Clark," I ask, "I'm trying to figure out just what is bothering you. Is it that Collin has suppressed causation in his account of explanation? Is the sticking point that explanation has to be causal, the revealing of causes?"
"No, that's not it." Clark responded as best as I can now recall, "I'm just asking if you deleted explanation from your account of the models, what would be lost?"
The wheels spun in my head and tumblers fell into place.
"Oh, I get it, you just don't like explanation period. It's nothing to do with causes!"
The moment of enlightenment was satisfying and led immediately to my next question:
"Why didn't you say it?!"
The comeback was a memorable moment in Center repartee.
"I did say it. Why didn't you hear it?!"
That was the beginning of a long, energetic question time. It's hard for me now to identify all the strands pursued. One major one, however, was Collin's excision of causation from the explanations. That did not sit well in the room.
As this question came back in many forms, Collin began to concede that he could weaken the excision without losing anything that mattered. He was precluding causal processes in the sense of Salmon from his account. But he was allowing them in the Woodward interventionist sense. That was hard to resist since Woodward himself had just made the case for Collin's models employing explanation in the Woodward sense.
Collin finally produced a good summary. "It's not about what is causal and non-causal. It's about what is explanatory." I thought that said it well.
The minutes where ticking down and it was time to hurry things along. "We have three questions in the queue and nine minutes. That's three minutes a piece."
Then: "We have two minutes. Edouard then Julia..."
The gallant Frenchman gestured that Julia should go first. Clark's finger was in the air and he was saying "I need ten seconds."
While Julia spoke, I retrieved the umbrella and was holding it visibly. The tension was rising. It was beginning to feel like the closing moments of an auction run amok. All might be won or lost with the wave of a hand.
"We have thirty seconds..."
"Then," said Edouard firmly before Clark could start, "I will take them." This new bid roused a chorus of amused murmurs.
After Edouard's thirty seconds, Clark decided that he now needed only five seconds. I yielded to the inevitable. Allocating 1.25 seconds to each word, he said:
"Explaining is an honorific."
The last words had been said. The applause was spontaneous and Collin received his umbrella. He had earned it.
John D. Norton