As an aged modeller (JK but I've been doing this for 10+ years LMAOSOB), I'm gonna share some basics I think everybody, ESP JUNIOR PEOPLE, in cognitive science, psychology & neuroscience should know!
Also hi this's what modellers (I & @andrea_e_martin) can look like!



1/
Also hi this's what modellers (I & @andrea_e_martin) can look like!



1/
1. Even a 100% flawed model can be useful.
So even a model that is unhinged from empirical evidence can be useful in that it can highlight what it is we do actually believe & what "unhinged" means.
We talk about it a bit here as well & cite examples: https://doi.org/10.31234/osf.io/rybh9
2/
So even a model that is unhinged from empirical evidence can be useful in that it can highlight what it is we do actually believe & what "unhinged" means.
We talk about it a bit here as well & cite examples: https://doi.org/10.31234/osf.io/rybh9
2/
2. Nobody/no modeller understands a totally novel to us modelling methodology by just reading once the first paper we have seen that uses it. We read papers many times and read many related models.
If I get a model immediately, it's because I have seen similar stuff before.
3/
If I get a model immediately, it's because I have seen similar stuff before.
3/
3. Models, like theories, are most likely to be kinda wrong & it's our job to admit that &, if we so wish & in an ideal world, refine them or get rid of them. People who behave otherwise are disingenuous & being megalomaniacal about their body of work.
You have been warned.
4/
You have been warned.
4/
4. There is bad modelling. In the same way work that avoids modelling avoids theory building, as we say in: https://doi.org/10.31234/osf.io/rybh9 — a lot of modelling avoids spec writing, as we say in: https://doi.org/10.1016/j.cogsys.2013.05.001!
5/
5/
It is scientifically dubious, even tho sadly common in modelling papers to not write a spec & just talk about code/superficial implementation details. It's nigh on useless, i.e., bad modelling and theory building, to do this and arguably lazy. See: https://doi.org/10.1016/j.cogsys.2013.05.001
6/
6/
What @andrea_e_martin and I propose is that we as (cognitive) scientists have to engage with every step in path our model. That doesn't mean everybody needs to be a modeller or that every modeller needs to collect data, or anything THIS dramatic...
7/
7/
...just that since science is the process of building theories of the world, we should strive to do just that.
More here:
7/
More here:
7/