Models

The blog is cause, models and inference: developing a knowledge based practice. On January 4th I said I would attempt to more clearly articulate the purpose of the blog by first defining terms. The past two posts have been about cause. We saw that defining cause was rather easy, but that philosophers have debated (since Hume) whether it is possible to know that a causal relationship exists from empirical evidence, and since Hume was an empiricist he meant by observations alone. The constant conjunction definition he proposed was an attempt to acknowledge what we do - we observe, we make causal inferences, we then manipulate causes to obtain desired effects - while remaining an empiricist. The fact that cause is important for clinical practice is clear (though I most likely will make another post on that once I am past all the defining and the establishing the foundation).

Now we turn to models. Anyone that has had a class with me is already familiar with my use of models in the development of knowledge. I take the grown (sigh, angst) rising from the students when I go to the blackboard with chalk in hand to crudely sketch a series of shapes, words and arrows as a clear sign that I am teaching them. As Rhett Allain has said: “Confusion is the sweat of learning” (“Why giving you the answer isn’t the answer”: here, for some great graphics: here), so if my chalk board sketches initially confuse students, when they come to understanding the “model” on the board then they have learned something. They now understand a model of reality. They understand something we have rationally built (in our minds) based on empirical observation that are subjected to testing through further observation. And that they can manipulate for desired effects.

On this blog I am attempting to develop a system that:

(Critical Realism) -> ((Empirical observations of real world + Critical mental reflection & model building) = causal models)) -> Causal Inferences -> Knowledge based practice

There - that is a model. Where does “evidence” fit in the traditional EBP perspective - technically simply the “empirical observations of the real world.” Traditional EBP puts more emphasis on the empirical observations than the mental models for making clinical decisions. I do not think that is what people ever really meant by EBP though. I think what I am proposing is more in line with the intentions of EBP, it is just not clearly articulated - but now I am getting off track. A later post will address the implications of aligning intentions with the words we use and our system of thought.

Models are mental representations of causal associations of a system (causal networks, Bayesian networks, structural models), based on empirical observation but involve a critical mental component and neither the empirical or the rationale alone makes the knowledge that can then be used. It turns out I do not have a need to get into too much detail here, now, but it is coming once we get from the foundations of a KBP to trying to demonstrate its implications. At that time the work of Pearl becomes very important.

Of course, as George Box proposed:  ”All models are wrong, some models are useful”

No model can account for the full causal network (structure) - across all scales (strata) of reality and time.  So they are all wrong in that sense. They are rationale (critical) abstractions of something out there that is real (realism).

Now this leads us into a bit of a mire regarding what we can know. If, after all, our models represent the balance of empirical and rationale, but the models are wrong but useful, does that leave pragmatism as the only option for determining what is true? I say no, but, given where I am headed (knowledge based practice for physical therapy specifically, health care generally) pragmatism is not a bad place to stop.

There is the “pragmatic theory of truth” particularly associated with William James, that “the truth of a statement can be defined in terms of the utility of accepting it.” (ODP) This I do not accept, I believe in truth separate from the utility of it. To do that I need to balance empiricism and rationalism with a presupposition (any well balanced stool has at least 3 legs) of a created order underlying all of reality that allows empiricism and rationalism to approximate truth, but for which truth ultimately has to be revealed to be known.

But “pragmatism” associated with Peirce does not proclaim utility as an arbitrator of truth, but that utility does provide meaning. “…the core is the belief that the meaning of a doctrine is the same as the practical effects of adopting it. Peirce interpreted a theoretical sentence [model] as a confused form of thought [all models are wrong] whose meaning is only that of a corresponding practical maxim (telling us what to do in some circumstance)[but some models are useful].” (ODP) Note that the [bracketed] words are my additions to the ODP quote. This is an acceptable mode of arbitration for a knowledge based practice, and as you can see, models are an important component. This sort of arbitration based on a practical maxim will be helpful to us when we start figuring how to balance the “empirical and rationale”, the “observations and the theory” in a knowledge based practice. If you need to be reminded of the importance of an arbiter of the balance between empiricism and rationalism you need only go back to yesterday’s post (January 6, 2014).

Moving on for today - the next post will take on the last of the primary terms - inference.

Please post comments, or send me an email (sean_collins@uml.edu) if there are sticky points that need further explanation. Also, please share with any colleagues or friends that you believe would be interested in this topic.

Leave a Comment