Why disagree?

Adrian Currie writes...

Homo floresiensis were a species of hominid which survived long enough to (temporally, if not spatially) overlap with modern humans. Since their discovery around a decade ago, paleoanthropologists have argued over whether they represented the last surviving lineage of a hitherto unknown radiation out of Africa, or whether they are dwarfed representatives of a later—and known—radiation. Recently, I asked whether new phylogenetic evidence was sufficient to conclude in favour of the former hypothesis over the latter. In this post, I want to ask a quite different question about the debate: whether it has been ‘productive’, and why. And relatedly, whether ‘disagreement’ is something we want scientists to engage in.

I’ll start with some philosophical scene-setting.

Presumably for a bunch of contingent, historical and socio-political reasons, philosophers have often been focused on—let’s call it—individual-level epistemology. When we’re considering scientific arguments, episodes, or debates, we tend to ask things like do the reasons provided ground justified belief? We ask, in a sense, what some suitably idealized rational agent would think or believe in light of the evidence and arguments presented. And presumably this matters: we want science to at least sometimes help us decide what propositions to commit to, what policies to adopt, and so forth. But that’s not all there is for us philosophers to wonder about. And for me (presumably for a bunch of contingent, historical, psychological reasons) it isn’t the most interesting thing for us to wonder about. In trying to understand paleontology, I’ve found myself more and more attracted to viewing science as primarily being a social process (um, I’m not particularly ground-breaking in that respect!): that is, we should understand scientific knowledge primarily in terms of the social practices which generate it. Specifically for this post, I’ve begun wondering about what I’ve been calling ‘productivity’.

The productivity of a scientific community is its tendency to produce bountiful, and diverse, ‘epistemic goods’. Think about an epistemic good as some piece of knowledge, where ‘knowledge’ is understood in a broad, pluralistic way. Scientific studies produce new technologies, speculative hypotheses, explanations, techniques, theoretical understanding, evidence, data, treatments, phenomena, syntheses, narratives, and so on and so forth. It might be that some features of scientific communities make them more or less productive, and identifying those features looks pretty important, I reckon. In a later post I’ll do some philosophical jiujutsu and put some clothes on this notion, but for now a common-sense description will do.

As I’ve mentioned, there’s been a debate in paleoanthropology about the evolutionary history of H. floresiensis, and (despite my timely intervention last month!) so far as I can tell, consensus is starting to form around one hypothesis over another. It looks like we should think that the hobbits were more closely related to Homo habilus than Homo erectus. That’s an exciting result, as it dramatically reshapes our picture of hominid dispersal out of Africa over the last 2 million or so years.

Homo floresiensis and some of the odd fauna she shared her environment with (from the National Museum of Nature and Science, Tokyo).

Homo floresiensis and some of the odd fauna she shared her environment with (from the National Museum of Nature and Science, Tokyo).

Now, considering the debate from individual-level epistemology—and assuming my scepticism last month can be waved aside—we might conclude that, say, the erectine hypothesis failed. It is likely false, and so should be put aside. We might also take the forming consensus as good evidence for success in this part of paleoanthropology, and indeed we might (as Finnur Delson recently argued in the context of climate change) take the remaining disagreement to signal a trustworthy epistemic community. But what about my question about whether disagreement is productive. That is, is it something that makes scientific communities more likely to provide epistemic goods?

From the perspective I outlined above, where we care about the ‘productivity’ of scientific communities, what is interesting about disagreement? I can put this as a kind of challenge: does disagreement give us anything more than diversity? Many people think that a diverse science is good: there is something excellent about a diversity of views, methods, and so forth, in science. But if disagreement just lines up with diversity, then it’s not bringing anything new to the party.

Here’s a way of putting things. Consider a community of, let’s call them, “Hasok-scientists”. Hasok Chang has suggested that consensus is over-valued in science, and that historical cases of consensus (such as the oxygen theory over the phlogiston theory in chemistry) were actually mistakes.  Hasok scientists don’t actively disagree. They have different views, etc… but are pluralists rather than outright disagree-ers. Let a thousand flowers bloom, they often announce, shrugging their shoulders at other’s views. Such a community is diverse, but it doesn’t contain active disagreement. Indeed, there are some reasons to think such a community might be better, as it could avoid converging too quickly on consensus, and maintain its productive diversity for longer.

A crack team of Hasok-scientists hard at work.

A crack team of Hasok-scientists hard at work.

Our question, then, is whether there are ways in which a population of actively disagreeing scientists might outperform Hasok-scientists. By looking at the H. floresiensis debate I want to suggest that, yes, there is a sense in which active disagreement might matter

What, then, might actual disagreement bring to a community: can it be productive? And moreoever, what should we say about the productivity of failed hypotheses—which we’ll take the erectine hypothesis to be for today’s post? As we’ll see, the erectine hypothesis was extremely productive insofar as testing it involved a very specific investigation of how phyletic dwarfism both evolves and is expressed in development in mammals, primates in particular. I’ll use the case to suggest a context in which disagreement is productive.

As we’ve seen, a crucial question concerning H. floresiensis’ ancestry is whether it is plausible that phyletic dwarfism in an erectine lineage would generate the hobbitses' distinctive set of traits. There’s initial reason, in fact, to think that it can’t.

Back in 2006, Martin et al argued that because of how dwarfism is expressed in development, the erectine hypothesis cannot explain H. floresiensis’ traits. The idea is that you evolve into a dwarf by cutting off your growing period early. This means that things which develop earlier should have had more time to grow than things which develop later. In mammals, brains are early developers, while overall body size spends longer growing. The effect of cutting off growth, then, should be critters with larger brains relative to their body size (that is, a higher ‘encephalization quotient’ or ‘EQ’) than their non-dwarfed cousins. The argument can be summarized:

(1)    Dwarfism is expressed developmentally by shorter growth periods;

(2)    Brain development completes earlier than body development in ontogeny;

(3)    So, dwarf species should have higher EQ.

Based on their developmental model, Martin et al predicted an encaphalization for H. floresiensis twice as high as it in fact was. Convincing evidence, they argued, against the erectine hypothesis. However, it was soon noted that Martin et al’s developmental model didn’t simply apply to potentially dwarfed hominids, but to a range of mammals who we have independent reason to think are dwarfed.

In 2009, Weston & Lister tested Martin et al’s model by comparing extant and extinct dwarf elephants and hippos to their bulkier cousins. They found that the model also fouled up for our distant relations in the ungulates. This suggests that island dwarfism could be playing a role after all:

Whatever the explanation for the tiny brain of H. floresiensis relative to body size, the evidence presented here suggests that the phenomenon of insular dwarfism could have played a part in its evolution.

Of course, this doesn’t provide much positive evidence that it does in fact play such a role. Earlier than Weston & Lister, and in part in response to questions arising from H. floresiensis’ ancestry, Bromham and Cardillo (2007) surveyed a bunch of primates to test whether the ‘island rule’—whether island-dwelling lineages were smaller in terms of body size than mainlanders—and found that it holds. However, in 2013 Stephen Montgomery adapted Bromham and Cardillo’s data to test whether such a rule holds for EQ. That is, he asked whether there are any clear patterns concerning relative brain and body size across closely related primates on islands and mainlands. It turns out there wasn’t much to find:

For EQ in particular there is little hint of any consistent effect of insularity with increases being as common as decreases. The scaling components between brain and body mass vary greatly across pairs from 0.24 to 1.24. This variability in response is a likely cause of the lack of an ‘island rule’ for relative brain size (756).

So, the effects of island dwarfism (‘insularity’) doesn’t have easy correlations in terms of shifts in brain size proportional to body size across primates. Regardless, Montgomery is able to construct a little inference tool which tests whether a particular discrepancy between two EQs could be explained by insularity within the range of primates. Sadly for the erectine hypothesis, it looks as if even by this test, it is very unlikely that Homo floresiensis could have evolved via phyletic dwarfism from an erectine hypothesis.

So, one thing we can probably conclude is that my previous attempt to remain optimistic about the truth of the erectine hypothesis cannot be upheld. Not only is there increasing evidence that there was a radiation earlier than H. erectus, but our more sophisticated understanding of island dwarfism doesn’t uphold the idea that the hobbits were erectine dwarves. However, I want to say that my optimism was not misplaced regarding the productivity of the hypothesis, and that this little history has important lessons pertaining to the question at hand: whether disagreement can lead to productivity.

Recall my question: is there something that a population of actively disagreeing-scientists can achieve which a group of Hasok-scientists would not? I’m inclined to say that in this circumstance there is something interesting going on. Although the erectine hypothesis is looking pretty unlikely, the road to learning this involved learning a bunch of other things which, to my mind, are even more interesting than the main target of the investigation (of course, philosophers have infamously odd ideas about what is ‘interesting’). We now have a much richer understanding of the nature of the evolution and ontogenetic expression of phyletic dwarfism across mammals and primates, as well as the capacity to develop local models which test the island rule for various traits in various lineages. This research was motivated, it seems to me, by the back-and-forth between those who supported the erectine hypothesis, and those supporting the habiline hypothesis. Potentially, this rich new knowledge was a result of the direct disagreement about the ancestry of this single lineage.

Disagreement (as opposed to pluralism), then, looks like it’s something we might want in our epistemic communities – but I suspect only under certain conditions. Testing the erectine hypothesis opened a fairly direct and clear line of argument concerning island dwarfism, and motivated a fairly concentrated set of studies regarding them. It strikes me that many scientific disagreements don’t have that character and here we might really want diversity, rather than disagreement itself.