Full Program »
Parameterization as a Framework for Modeling Contested Scientific ConceptsMany scientific concepts are contested in the sense that experts disagree about their proper use and extension. For example, biologists employ competing notions of SPECIES, GENE, and NATURAL SELECTION, and cognitive scientists disagree about the nature and extension of COGNITION and REPRESENTATION. However, although we have several well-established strategies for modeling scientific and lay concepts—e.g. with definitions, or homeostatic property clusters—we have few illuminating methods for modeling contested concepts. I propose a method called parameterization for fruitfully modeling these concepts. I discuss its benefits, and illustrate it with two examples.
In a parameterized explication of a concept, some terms are treated as parameters that can have various interpretations as values. By varying the interpretation of a parameter we can model variations in the extension of the explicated term, thus replicating the diversity of ascriptions and inferential roles we find in scientific practice. A successful parameterization reveals both a core of common practices concerning the use of a term (in its unparameterized elements), as well as major topics of disagreement (in its parameterized elements). These topics of disagreement are often ripe for further empirical or philosophical inquiry.
For example, the case of NATURAL SELECTION offers an example of relatively well-covered territory. Consider the following account of selection, based on Richard Lewontin’s account:
Selection occurs where:
1. there is a population of INDIVIDUALS,
2. there is variation in the traits of those individuals,
3. this variation causes variation in the FITNESS of individuals,
4. possession of the traits is partly HERITABLE.
Let the capitalized expressions be parameters, whose precise interpretation may vary. By accepting interpretations of the parameters that have different extensions, selectional phenomena can selectively include or exclude boundary cases such as breeding (artificial selection), some forms of cultural evolution, and genetic algorithms. For example, in breeding and evolutionary algorithms, but not in paradigm cases of natural selection, the fitness function is largely determined by human decision procedures. In evolutionary algorithms, cultural evolution, and viral evolution, heritability does not depend on germ-line transmission. The parameterized account makes explicit the dimensions of flexibility in the concept of SELECTION that enable its generalizability to novel contexts.
The scientific concept of COGNITION, by contrast, is not as well-understood. Cognitive scientists engage in disputes about whether cognition happens outside of brains—in the extra-neural body, outside the body, or in plants or microbes. Extant debates about cognition tend to turn on how to interpret “representation,” glossing over other issues. The following explication models the variance in scientists’ judgments:
Cognition is the operation of mechanisms, where
1. the mechanisms BELONG TO A SUBJECT,
2. the mechanisms REPRESENT the subject’s environment, and
3. the mechanisms manage the BEHAVIOR of the subject.
If the explication above is adequate to the variation in scientific judgments about the extension of COGNITION, it suggests that further philosophical attention to the nature of cognitive subjects and behavior is warranted.
Thus parameterized models offer perspicuous representations encapsulating deeper understandings of contested scientific concepts than simpler alternatives.
John V. Roach Honors College
Texas Christian University