What is a Scientific Model ?

[studiot]

A very popular phrase in this website is in the words of Dr Swanson.

Quote

You don’t have a model.

Another member's longer description of a model is also quoted here

Quote

A scientific model is the representation of a phenomenon, rather than a verbal explanation. You could use visual models like charts, or a computer simulation, for example. Building a physical model is difficult at the levels you're discussing, but that's another example of a model. Using maths to represent how the phenomenon is calculated is the perfect tool at the quantum level. Without any of these, you don't have a model.

 

The purpose of this thread is to arrive at a robust definition/description of a scientific model by discussion.

Perhaps a summary could then be pinned for future reference as the same issues come up time and time again.

 

To kick off I note that using just model by itself can be unhelpfully vague and there are many descriptors/qualifiers available.

Models can be ( in no partuicular order) qualitative; quantitative; illustrative; comparative; graphical; analogue; digital; expressed by a process; expressed by a formula and so on.

Since no model completely describes the subject, additional information is needed about the actual qualities described and any limitations imposed.

[iNow]

8 minutes ago, studiot said:

Models can be ( in no partuicular order) qualitative; quantitative; illustrative; comparative; graphical; analogue; digital; expressed by a process; expressed by a formula and so on.

Models accurately explain current behavior and make testable predictions about future behaviors observed in specific circumstances. It's not enough to say that X or Y will change, but instead must be able to forecast the outcome as Z.

[Phi for All]

3 minutes ago, iNow said:

Models accurately explain current behavior and make testable predictions about future behaviors observed in specific circumstances. It's not enough to say that X or Y will change, but instead must be able to forecast the outcome as Z.

This is all true, but this also describes what a theory is doing. Models need to represent a phenomenon, to show what the explanation is talking about. They should provide a valid approximation of the behavior as a perspective on the phenomenon. Like Eddington using spherical cows to illustrate a particular aspect or insight.

[studiot]

1 hour ago, iNow said:

Models accurately explain current behavior and make testable predictions about future behaviors observed in specific circumstances. It's not enough to say that X or Y will change, but instead must be able to forecast the outcome as Z.

What about Newton's 3rd Law ?

[Genady]

This physics professor describes his hierarchy of Framework >> Theory >> Model. Theory is a framework applied to a specific context, model is a theory with external parameters plugged in. Might be useful for this discussion. See the lecture between 14 and 22 minutes: 

 

[iNow]

5 hours ago, studiot said:

What about Newton's 3rd Law ?

What about it? It clearly describes outcome Z, even if imprecisely. 

[StringJunky]

A scientific model, to me, is an abstract construction, usually mathematical, that describes relationships between phenomena and objects. Models don't attempt to describe the total phenomenology, but some specific aspect of them.

Edited January 24, 2022 by StringJunky

[studiot]

2 hours ago, iNow said:

What about it? It clearly describes outcome Z, even if imprecisely. 

So you consider N3 as a model ?

I rather read a greater requirement into your original answer.

Below are two models.

Do either model N3 in your opinion ?

 

1 hour ago, StringJunky said:

A scientific model, to me, is an abstract construction, usually mathematical, that describes relationships between phenomena and objects. Models don't attempt to describe the total phenomenology, but some specific aspect of them.

How about models A and B below ?

Do they meet your standards ?

 

[iNow]

I don’t feel like quibbling, but overall yes. I’d be comfortable calling those models. YMMV

[StringJunky]

1 hour ago, studiot said:

So you consider N3 as a model ?

I rather read a greater requirement into your original answer.

Below are two models.

Do either model N3 in your opinion ?

 

How about models A and B below ?

Do they meet your standards ?

 

Yes, it's a schematic model description of something.

[OldChemE]

Well, OK-- one attempt:  A Model is an artificial construct, often mathematical, which represents a real world phenomenon, and which predicts the real-world outcome that is expected to result from a real-world input.