Chapter 5 Flashcards
Modelling
Representation
Something is a representation of something else if it stands in for that thing, meaning we use or investigate that thing instead of what we are really interested in
Target
A target is that which a model aims to represent
Idealization
The fact that a model is an idealization of its target means that it does not have all the properties that the target has, and vice versa
Purpose dependence
A model can only be justified on the basis of how useful it is for fulfilling a certain purpose
Analogy
Inference by analogy is to state that a conclusion in one case applies to another case due to there being relevant similarities between the cases
Positive analogy
Similarity between model and target
Negative analogy
Dissimilarity between model and target
Neutral analogy
Property of model, whose occurrence in target is not known
Epistemic virtues of models
Types of properties of models that determine the quality of a model, given a specific purpose
Similarity
The degree to which the model has several properties which the target also has (there are multiple positive analogies) that are relevant for the modelling purpose
Robustness
A model is robust with respect to an assumption if changing this assumption does not change the model result
Parameter precision
One model has higher parameter precision than another model, if the specifications of the parameters of the first model implies the parameter specifications of the second
Simplicity
One model is simpler than another model if it contains less parameters than another model
Tractability
A model is computationally tractable only if its result can be computed in polynomial time. A model is analytically tractable only if the model result can be obtained through valid deduction, from the model assumptions alone. A model is theoretically tractable only if the model is either computationally or analytically tractable, and some of the necessary computational or deductive steps are justified with reference to a background theory
Transparency
A model is epistemically transparent if the model user is cognitively capable of understanding how the model result is produced