Are there ecological argument forms?
The second reflection that Betts et al.’s argument structure raises for
me is especially pertinent to ecology as a science. I am struck that the
form of argument that Betts et al. present, in which the scientist is a
rational actor passively selected on by his environment, is drawn not
from ecology, but from evolutionary biology. If there were a standard
argument form about, say, ecological succession that the authors had
drawn on, using it here could be a clever and ironic meta-argument (if
still not well adapted to the question at hand). However, it appears to
me that there are no standard forms of argument about ecological
phenomena. The closest thing I can think of is the idea that you derive
hypotheses by extending a mechanism observed at a small scale across
time and space. However, the qualities of various forms of reasoning
about pattern interactions or up- and down-scalings over time and space
are not well established. Ecology has systematically undervalued all
forms of non-statistical reasoning (statistical reasoning also requires
further discussion, e.g. Amrhein et al. 2019). Reasoning across time and
space is, however, about more than statistics. It should also involve
judgements about when, where and why certain kinds of scaling phenomena
occur.
For example, we often use emergent-phenomena reasoning to explain the
interactions of monospecific groups, but rarely heterospecific groups
and never the assembly of species communities—why? We often use
attractor states to explain degradation, but not to explain community
assembly or interactions of monspecific groups—why? If I were to
propose that we model degradation as being caused by an emergent
property of multiple species interactions, or that we think of termite
mounds as formed by moving towards structural attractors, these might
sound “wrong”, but I posit that ecologists have developed very few
logical or rhetorical resources for arguing for or against such
proposals. Some ecologists might object here that the argument behind
the model is irrelevant because the model either corresponds to (or
predicts) empirical observations, or not. However, I am willing to bet
that by relabelling the variables in a model of termite nest building or
a model of degradation, the phenomenon they are supposed to be
representing could be switched such that the model of degradation would
now explain the canalization of the nest shape into a typical form,
while the model of the termite nest would explain the reordering of
species into an irreversible pattern. It is thus the model that is
relatively irrelevant—only the argument specifies what it is that we
claim to be explaining, and thus how we evaluate the model.
Betts et al., like most ecologists, rely entirely on statistical
reasoning as their analytical approch to answering their questions, and
thus structure the core of their argument in the form of predictions
about whether hypothesis use has increased or decreased. This is,
frankly, the most boring possible question that could be asked about
hypothesis use. Other things one would like ask questions about are, at
least in my opinion, the things I am talking about here. However, the
question of whether hypothesis use has gone up or down is also, I
recognise, one of the only questions that can be tested using available
quantitative data and formulated as something that can be answered ‘yes’
or ‘no’ through statistics. Betts et al. thus provides a demonstration
of how a focus on hypothetico-deductive reasoning can limit researchers
to asking and answering uninteresting, overly simplifed questions.