Nersessian, N. J. (1998). Model-based reasoning in conceptual change. in Model-Based Reasoning in Scientific Discovery, L. Magnani, N. Nersessian & P. Thagard (eds). Plenum.

@InBook{,
  ALTauthor = 	 {Nancy J. Nersessian},
  ALTeditor = 	 {Lorenzo Magnani, Nancy J. Nersessian, Paul Thagard},
  title = 	 {Model-based reasoning in conceptual change},
  chapter = 	 {},
  publisher = 	 {Plenum},
  year = 	 {1999},
  OPTpages = 	 {},
}

Author of the summary: Jim R. Davies, 2000, jim@jimdavies.org

Cite this paper for:

• traditional philosophy says that reasoning is application of deductive or inductive algorithms to propositions [p2]
• mental modeling: thought experimentation on mental models [p4]
• Mental modeling capacity may have evolved so creatures could simulate moving in the environment
• Conceptual change is not a sudden insight, but a result of extended problem-solving practices. [p7]
• generic abstraction
• Scientists may use visual representations because they lack the constraints inherent in formulaic and linguistic ones. [p10]

In contrast, Nancy's approach looks at the processes that produce these new concepts.

3 practices that generate conceptual change through model-based reasoning are the focus of this study:

1. creating analogies
2. employing visual representations
3. thought experimentation

Model: interpretations of the processes, phenomena, and situations of physical systems. [p5]

The traditional philosophical view of reasoning does not cover the generative forms of reasoning in this paper, which are not guaranteed to produce an adequate solution.

These claims apply to reasoning in physics, and may apply to other domains because they are based on fundamental cognition.

The main argument of this paper: [p2]

1. There is a cognitive foundation for modeling practices as productive forms of reasoning
2. They are more general and applicable then in specific scientific contexts
3. therefore they are "extremely productive forms of reasoning in conceptual change in science"

Experiments such as the Wason card task show that semantic information is important to human reasoning, so it's not just logic. This led to the "mental modeling" view (Craik 1943). This is thought experiment on mental models. Mental models theory had been argued to provide a unifying framework for cognition (Gilhooly 1986). [p10]

Nancy's mental model theory is in some ways an intellectual extension of Johnson-Laird's account (1983). [p5] Mental modeling capacity may have evolved so creatures could simulate moving in the environment for planning purposes.

Conceptual change is not a sudden insight, but a result of extended problem-solving practices. [p7] Models are the basic unit of scientific reasoning is the model (as opposed to the proposition, e.g.). [p8] Models change first, and from them we get laws and axioms of theories.

Newton could only have reasoned about commonalities among the motions of planets and projectiles by the process of generic abstraction.

Scientists may use visual representations because they lack the constraints inherent in formulaic and linguistic ones. [p10] In this work models are analog and non-propositional.[p12]

Bohr claimed he could not visualize well.[p12]

Maxwell's vortex Figure with idlewheels is "a visual representation of an analogical model that is accompanied ith instructions for animating it correctly in thought." [p13]

Conceptual change results from extended problem-solving practices. [p14]

Often in examples of conceptual change, no direct analogy exists and a source analog must be constructed. [p15]

You need to do generic modeling to be able to do analogy across domains. Generic models are representations of features common to a class of phenomena. [p16]

For Maxwell, "the vortex-idle wheel system is understood to represent the class of such dynamical systems and the class includes electric and magnetic interactions on the assumptions of Maxwell's model."

Thought experimenting is the running of mental models.[p19]

Summary author's notes:

questions and confusions:
• Mental models can be represented propositionally, right?
• This paper makes no claim that maxwell used the diagram or anything like it to generate the theory.
• What is the evidence that there is often no direct analogy to use for a source, so that you have to use an abstraction or a generic model?
• Did maxwell use a generic model to retrieve the idle wheel solution from gear systems? If so, how did he create the generic model? From lots of gears? If so, how was it identified as a member of that class of systems? If the generic model came from the vortices, then what was the class of phenomena he used to find common features to make the abstraction?