Wayne's Razor: in defence of speculation

Adrian Currie writes...



I’ll get to the outdated pop-culture references soon, but let’s start with some Tyrannosaurus rex facts (taken in part from Currie 1998 and Erickson et al 2004, as well as the controversial ‘Dino Gangs’ documentary).

First, there’s some reason to think T. rex were clever: the brain case is pretty big for an animal of that size, compared with modern reptilian mega-carnivores (crocodiles and their ilk). Second, there’s reason to think they lived in groups, as fossil finds containing multiple T. rex, of multiple ages, have been found (as have multiple trackways). Third, there’s some reason to think T. rex physiology changed a fair bit with age. The young ones ran perhaps twice adult speed, but it’s the adults who have the great big, killing bites. Oh, and here’s a fourth bit of information: T. rex prey, larger herbivorous dinosaurs, could probably run faster than a fully grown adult T. rex.

Okay, so what? In a documentary around 5 years ago Phil Currie (no relation!) suggested this tells us that T. rex weren’t just gregarious, but hunted in packs. That is, they actively collaborated in taking down prey. The younger, swifter T. rex would go first, catching and harrying their intended victims—slowing them down—enabling the adults to move in for the kill, delivered via their trademark crushing bite. It’s a terrifying thought: I like to imagine the elder T. rex treating the juveniles a bit like an upper-class buffoon treats his fox-hunting dogs.

When I present Currie’s hypothesis to non-specialist audiences with a straight face, they buy it. Which is weird, considering that (1) the idea was cooked up for a television show; (2) the reaction from the paleontological community ranged from a skeptically raised eyebrow to something approaching blistering rage—Brian Switek describing it as ‘nauseating hype’; (3) the evidence is, let’s face it, pretty rubbish. Let’s focus on that third point.

There’s no evidence that T. rex didn’t pack hunt. The big brain, the gregarious living, and so forth, suggest that the behavior is within the realms of possibility. But ‘within the realms of possibility’ is a pretty pathetic bar to set for scientific assertability. There are many, many possible explanations of our evidence. For example, T. rex might only meet at particular times, perhaps for mating purposes. Or they could (as Lawrence Witmer suggested) engage in ‘communal hunting’—that is, they could opportunistically share in kills, but not actually coordinate their behavior. Moreover, the overlapping finds could be due to some natural process: a river could have swept remains from different locations (although recent trackway discoveries undermine this). All kinds of past states of affairs could conspire to leave us with the T. rex remains that we have.

In short, the hypothesis that T. rex pack hunted is underdetermined: we just don’t have the empirical evidence to discriminate between it and other possibilities.

When facing underdetermined hypotheses, the obvious thing to do is look for ways of discriminating between them. One approach—and this is where Wayne comes in—is to find some virtue that one hypothesis has which the others don’t. Such virtues could act as tie-breakers. Philosophers of science discuss many such virtues: notions of simplicity, including Ockham’s razor (one shouldn’t multiply entities beyond necessity!), explanatory breadth, explanatory ‘goodness’, unity, and so forth.

These could be virtues because the hypotheses which bear them are more likely to be true. On this reading, by Ockham’s razor, if all else is equal, the hypothesis with the fewest types of entities is more likely to capture how things are. Why would we think that? We’d need some reason to think the world is as the virtue describes. Imagine we knew there was a creator god, and moreover knew that she really liked limiting how many types of entities she created. That would seem to justify taking Ockham’s razor as a good way of choosing between hypotheses. If we have reason to think the world is simple, then simplicity can be a guide in deciding which hypotheses are likely to be true. On this kind of view, such principles take us from a hypothesis’ possession of the virtue to its likely truth.

Something like Ockham’s razor isn’t appropriate for our T. rex case, unfortunately: I don’t see how Currie’s hypothesis might be simpler than the others. However, perhaps there’s another ‘razor’ that could do the trick:

Wayne’s Razor: for two hypotheses which are otherwise empirically equivalent, we should prefer the hypothesis which is the most excellent.

Two terms matter here. First, what we mean by ‘prefer’, and second—of course—what it takes to be ‘excellent’. Excellence, I take it, is a quality adjacent to awesomeness, and I doubt a rigorous analysis can be provided (although). But it is fair to say that pack-hunting Tyrannosaurus rex would be, in some important sense, both awesome and excellent. What should we mean by ‘prefer’? Well, we could read ‘prefer’ in the likely to be true sense. That is, being excellent could be a virtue in the terms I described above.

Often, when I hear of a new hypothesis, I find myself thinking ‘damn that’s cool’, and find myself immediately inclined to believe it: Wayne’s Razor at work.

Sadly, this looks like a very difficult principle to defend in these terms. What principle could connect a hypothesis’ being excellent to its likely truth? Often, the most excellent thing is not true—just reflect on your favourite set of atrocities from human history (unless we happen to live in the most excellent of all possible worlds, of course). I would really like Wayne’s Razor, in the sense discussed so far, to be justified, but sadly, it isn’t. ‘Excellence’ just isn’t truth tracking.

However, perhaps we’ve been reading ‘prefer’ all wrong. There are many reasons to prefer one hypothesis over another—and some of these don’t involve truth. I’m going to suggest something along these lines, and it’s going to involve saying a few things about how I think historical investigation most fruitfully proceeds.

As we’ve seen, historical scientists often find themselves facing underdetermination—their evidence is often sketchy, degraded, and so forth. Under such circumstances, I’ve noticed that scientists sometimes get conservative: they become extremely careful to not say anything that might outrun their evidence, often for fear of being ‘unscientific’. I think this is a big mistake.

Why? It’s often extremely difficult to know what evidence might be available prior to carrying out an investigation, and to know what the results of the investigation might be relevant to. Betting on the future of a paleontological investigation is like getting blood from a narwhal: it’s hard to know where to start. Paleontology is riddled with lucky finds and remarkable technological innovation which just couldn’t have been predicted beforehand. In such a situation, here’s something which really helps you see your way forward: articulate a hypothesis.

As soon as a hypothesis is on the table, people start exploring how it might be tested. Currie’s hypothesis leads to a fascinating set of questions. Just how can we tell the difference between pack hunting, and less cooperative activities, based on paleontological remains? Just what can brain size tell us about cooperative behaviour? What kinds of tracks or other preservable signs do cooperatively hunting animals (like lions) leave which are indicative of their hunting strategy? And so forth. Without the hypothesis actually being articulated, it is unlikely that we would examine the world in such a light.

This is to say that speculation—possibly even speculation as loose as Currie’s—often drives progress in paleontology. Articulating hypotheses gives scientists a target to chase, while without one they could be left floundering.

This suggests another way of defending speculative hypotheses (without having to believe them!). Some speculative hypotheses drive scientific progress, that is, they suggest further empirical investigation—field work, the development of new technologies, uncovering new lines of evidence, or articulating further hypotheses. Moreover, often such hypotheses are necessary for these new empirical avenues to open up. And it seems that this is a good basis for preferring hypotheses: after all, by spurring new research they lead to new epistemic achievements. The ‘goodness’ of a paleontological hypothesis, then, isn’t decided on its likely truth alone – we should also consider whether it leads us to new empirical tests, whether it opens new routes to understanding the world. And in this, speculative hypotheses often excel.

Speculation, then, can be excellent.

*See here for Derek's response!*

References & Further Reading

I’ve always had a soft spot for Paul Churchland’s discussion of super empirical virtues:

Churchland, P. M. (1985). The ontological status of observables: in praise of the superempirical virtues. Images of science, 35-47.

Alan Baker’s SEP article on simplicity is a good place to look for discussion of Ockham’s razor and related notions:


Our own Derek Turner has some nice discussions of underdetermination in historical science:

Turner, D. (2005). Local Underdetermination in Historical Science. Philosophy of Science, 72(1), 209-230.

Turner, D. (2007). Making Prehistory. Historical Science and the Scientific Realism Debate.

Here are the other references:

Currie, P. J. (1998). Possible evidence of gregarious behavior in tyrannosaurids. Gaia, 15, 271-277.

Erickson, G. M., Makovicky, P. J., Currie, P. J., Norell, M. A., Yerby, S. A., & Brochu, C. A. (2004). Gigantism and comparative life-history parameters of tyrannosaurid dinosaurs. Nature, 430(7001), 772-775.