Tuesday, November 29, 2011

Locke's argument against innate speculative principles

To: Julian P.
From: Geoffrey Klempner
Subject: Locke's argument against innate speculative principles
Date: 1 March 2006 11:55

Dear Julian,

Thank you for your email of 20 February, with your University of London essay in response to the question, 'Does Locke's Argument Destroy any Possibility of Innate Speculative Principles?'

I have attached the two papers which I gave at Prague College. These will be appearing in a future issue/ issues of Philosophy for Business. The College want to set up a Centre for Business Practice with the backing of the British Chamber of Commerce Czech Republic which would offer seminars as well as publishing materials on business ethics/ corporate responsibility.

I find the innateness debate very difficult. The problem is seeing what is at stake, once one moves beyond the critique of naive innatism.

One reason for thinking that Locke's argument fails to show conclusively that there cannot be innate speculative principles is Peter Carruthers' idea (in 'Human Knowledge and Human Nature') that a modified form of innatism can be reconciled with empiricism, by bringing in a version of Quinian naturalized epistemology. Evolution, not God, has provided us with a certain degree of 'hard wiring', whose status as 'knowledge' is 'non-accidental' in the sense required.

In order to cast doubt on Locke's argument, it is not necessary to prove that the evolutionary theory of knowledge true. It is sufficient that it is not vulnerable to same objections as the God theory. We can't expect evolution to provide us with houses (although it does provide snails with a shell). What one could or could not acquire through natural selection is a matter for empirical investigation, seeking as always the 'best explanation'. Chomsky's theory of innate grammatical rules would be a case in point. It is a matter for empirical investigation and philosophical debate whether or not transformational grammar provides the best explanation for human language learning capacities, or whether, on the contrary, the Davidsonian program of constructing a theory of meaning in first-order predicate logic is sufficient for this purpose.

The Chomsky-Davidson debate over linguistic knowledge seems not dissimilar to the Leibniz-Locke debate. In these terms, Leibniz is arguing that God needs to give us more than the bare capacity to reason logically, in order to have the basic tool set needed to acquire empirical knowledge. But what exactly is this extra element and how does it work?

Your excellent question about an intelligent computer raises a deep issue, which relates to the laws of logic. The original Pentium chip - or it might have been Pentium II - was famously discovered to have a fatal defect which led to the wrong result for certain arithmetical computations. Macs, which I know better, use 'Math Libraries' in order to speed up calculations. A few years ago researchers at Motorola produced their own Math Library extension, MathLibMoto (there is also another version called LibMotoSh) which cuts corners in order to gain increased speed, at the cost of small errors in some calculations. In other words, whatever rules we build in, computers will apply accordingly. The correct rules will produce correct calculations and incorrect rules will produce incorrect calculations.

But we are talking about *intelligent* computers. We may define an intelligent computer as one which would be amenable to psychological explanation which posits beliefs and desires. Beliefs and desires are connected logically via something like the following formula:

It is a priori that for all A, p and x there are conditions C such that if x desires that p and believes that if he A's then p, and condition C obtains, then x A's.

(Taken from C. Peacocke 'Holistic Explanation' OUP 1978, p.11)

What this shows is that the grasp of principles of logic is built into the very concept of a psychological state, i.e. a state which can be accounted for by psychological explanation.

You say, 'We have to write the truth of the proposition [All F-things are F] into the program as a rule. But then we are just like God engraving innate speculative principles on the mind of the computer.' But the point is that, however such beings are 'produced', whether by God or evolution, they cannot fail to have what is required to recognize the laws of logic. Even God could not make an illogical intelligent being. Of course, human beings can appear 'illogical' at times, but this arises only against a background of adherence to the laws of logic.

In some way, the shift to considering the possibility of a priori knowledge misses the point of the innateness debate. Locke allows a priori knowledge but regards it as 'analytic'. But what is that? As you argue, there are obvious analytic truths and very far from obvious analytic truths. When does the ability to calculate become knowledge of the result of possible calculations which one has not actually performed? And what hangs on that question, anyway?

Like you, I suspect, I feel a residual sense of confusion regarding the innateness debate. What does the difference between Locke's final position and Leibniz amount to? Locke says we have what it takes to reason out innate speculative principles for ourselves. How does Leibniz's position differ? You say, 'If all necessary truths which we can grasp are analytical... Locke has won.' In order to establish a distinct position from Locke's, does Leibniz have to claim that (some) innate speculative principles are synthetic? I don't see this.

However, examples of putative 'synthetic' innate speculative principles would help.

All the best,