Saturday, February 23, 2019

Lars P. Syll — The limits of probabilistic reasoning

This is an important issue and some background is needed.

This is a fundamental issue in epistemology. As such it involves not only mathematics and science but also philosophy and logic. It is not an exaggeration to assert that this debate has been going on for millennia around the world and it remains undecided, which implies the need for further exploration. To claim certainty under such circumstances is premature.

Furthermore, the certainty of models comes from the logic and math. This certainty is that of tautology, or logical necessity, and contradiction, or logical impossibility. This is endogenous to the modeling process.

The interpretation of the model as a representation of reality is a substantial matter that goes beyond the procedure of the modeling and stands in need of connecting the model and reality in a way that is exogenous to the model. Process does not include substance as a virtue of the model. The model must be connected with what it represents internally, of course, and philosophy of logic and philosophy of science explore how this may take place. This is still controversial.

This controversy was raging at Cambridge and Oxford when Keynes was there. He would have been aware of the issues involved and the horizon of knowledge at the time. He was incorporating his view this in his Treatise on Probability.

In addition, at least since Aristotle, science has been regard as the search for causes and the aim is to provide a causal account that accords with observation. As Hume pointed out, the concept of causality is a very slippery one indeed. This is also a controversial subject and presently the debate in our area of the world is largely between realists and instrumentalists.

For example, accounting rules establish identities that all who understand the rules and their application accept. Where disagreement arises is often in attributing causality to the identity to account for it in actual terms and to use it for forecasting. This is to move from accounting to theory and scientific theories must be "testable" to distinguish them from speculation. The meaning and criteria of "testable" are also controversial.

Disagreement in debate often comes down to such matters, which are presumptions acting as hidden assumptions, since the underlying frame is not articulated and when parties disagree over the framing, there is no path to reaching a decisive outcome for lack of agreement over fundamental criteria.

Such issues regarding economics encompass philosophical logic (including foundations of mathematics), ontology, epistemology, value theory, action theory, philosophy of science and philosophy of social science. This is before getting into sociology and economic sociology, anthropology and economic anthropology, and history and economic history. Oh, I almost forget system theory and information theory.

In short, most conventional approaches are naïve unless they take foundations into account.

Lars P. Syll’s Blog
The limits of probabilistic reasoning
Lars P. Syll | Professor, Malmo University

See also from Lars today

Simplification is a virtue in modeling. Oversimplification is a vice.


Unknown said...

Paul Davidson had something useful to say on Keynes and probabilistic reasoning in connection with investment certainty:-

AXEC / E.K-H said...

Links on Lars Syll’s ‘The limits of probabilistic reasoning’

We simply do not know — so let us move on

Knowledge is attainable ― even in economics

Egmont Kakarot-Handtke

#Economics #FailedScience #FakeScience #CargoCultScience #ScientificIncompetence #Economists #MicroFoundations #MacroEconomics #Methodology #BehavioralEconomics #DeleteEconomics #DrainTheScientificSwamp #ParadigmShift #NewParadigm #Science #MacroFoundations

Noah Way said...

Probability that Egmont Kakarot-Handtke is a troll: 100%.