Monday 12 October 2020

Lecture 6: The Cartesian Program (Turvey, 2019, Lectures on Perception)

Turvey is arguing that modern psychology is still operating within a mechanistic framework, which assumes things like linear chains of causation and predicative properties. In the last lecture, he laid out the kinds of properties this framework allows, and connects his critique of them to the more interesting notions of causation and impredicativity he introduced in Lecture 4

In this Lecture, he spells out the way psychology implements the mechanistic framework and identifies that it is, specifically, a Cartesian implementation. Here, we will learn the form of the program, and note how it shows up in the familiar terms of cognitive psychology. 

Three Grades of Sense

The mechanistic hypothesis allows that one thing can cause another via a linear chain of cause and effect linked by the transmission of forces. This works okay for things such as 'how the rock broke the window' but immediately gets weird when you are trying to explain the behaviour of living organisms (Lecture 5). 

Descartes dodges part of this problem by first identifying a dualism, between humans and all other living organisms. He declares the latter to be completely mechanical, and so, while they may be complicated, a full account of their behaviour is possible just by applying the rules of a mechanistic science. He is still committed to a three term theory of perception (because of empirical findings such as the projection of an image by the lens onto the retina), but for animals all the terms are mechanical. 

This is his first grade of sense: reflexive behaviour brought about by the mechanical composition of the animal, with no awareness implied or required. Animals have this; humans do too, but clearly we also have something else. 

The second grade of sense is the fact that the mechanical activity in the first grade lead to basic awareness. For example, the mechanical activity of light on the retina comes with subjective experiences of sensations such as brightness. These are the secondary qualities we came across in the last lecture, the crossing from mechanical to mental qualities.

The third grade of sense is purely mental: it is the objective awareness of the world that results from the operation of rational processes working with the elements of the first and second grades. The goal of these processes is to identify the primary qualities that were the source of the secondary qualities - to identify the parts of the world. 

The Third Grade Requires Symbols and Operations

Descartes recognised that the capabilities of the third grade of sense required powerful machinery to work. As we've seen, 17th century thinkers were very impressed by the growing understanding of what mathematics allowed you to do with respect to knowing about the world. The notation system of mathematics is a set of abstract symbols, and their power comes from the fact they can be applied to almost anything (the number '4' might be a mass, a speed, a position, etc). The third grade of sense therefore works via the manipulation of symbols according to rules that constrain the moves to ones that remain coherent. 

The strength of symbols (the fact they are abstract and can therefore refer to nearly anything) is also a weakness. Given a symbol, how do we know to what it refers? This is essentially the grounding problem. Descartes 'solves' this by allowing for the existence of a set of innate knowledge that is sufficient to make the whole thing work (e.g. the fact that you shouldn't accept a contradiction such as something is both alive and dead at the same time). The third grade of sense applies these innate operations or operations derived from them to symbols, and the whole thing works because God provided the right set of innate knowledge. 

Loans of Intelligence

Descartes' analysis therefore has a fairly obvious problem, namely the appeal to a well-behaved God providing us with the right innate knowledge to ground everything else. In addition, there is a homunculus lurking too - who is doing the interpreting of the results of the operations? Dennett calls these kinds of problems 'loans of intelligence', where a theorist borrows what is required to get their theory up and running. These may be a legitimate temporary move, if used with care, but Descartes clearly had no intention of paying his loan back and it's not at all clear how to naturalise God in this account. 

It's worse than that, however - Descartes couldn't pay back this loan even if he wanted to. The language of the third grade of sense is intentional, in that it is 'about' something. We talk about 'decides [that]' or 'recognises [that]' or 'knows [that]'. You cannot translate such statements into the extensional language of the mechanistic hypothesis; they don't obey the same rules. This makes the Cartesian program founded on an unpayable loan of intelligence. 

The Cartesian Program

Despite the flaws, it's possible to identify the Cartesian program of research, and to see immediately that it is the program followed by modern psychology.
  1. Develop a complete mechanical account of the first grade of sense
  2. Identify the gaps between that account and what the second and third grade of sense can do (i.e. identify the required knowledge)
  3. Identify the symbol manipulation processes that implement that knowledge so as to make the third grade of sense possible.
This is literally what a representational program is doing. Modern psychology is a Cartesian program grounded in the mechanistic hypothesis. It therefore inherits all the problems and our evolving ecological solution has to therefore do things completely differently (as we have begun to sketch out). 


I'm really starting to see the shape of the book, now. Remember, it's called Lectures on Perception. Turvey is therefore laying out the history of theories of perception, specifically what motivated them and the rules they were following. Along the way, he has identified a bunch of problems, and begun to sprinkle possible solutions into the mix. Because of the nature of the problem he is describing, Turvey can't just lay out a nice linear story in which we progress along a nice pathway. That isn't going to be what happens - psychology is going to get stuck and start looping through various versions of the Cartesian program, suffering the same problems and never solving them. 

Funnily enough, I think this book is going to be impredicative - specifically, the package is going to make sense once it is all in place. This is just intrinsic to the topic, I think - I often struggle to describe the loopy, dynamical things I am up to using linear, sequential language, and I think this is why I've been finding this book hard. Turvey is embracing the impredicativity, but that makes the sense of progression a textbook normally gives you very loopy and hard to see. Writing these posts has helped me see, though.

I am entirely on board with the idea that psychology is stuck in a loop defined by the rules of a single program. My PhD student Danny Leach is writing up his dissertation about transfer of learning. His literature review/historical survey at the start is literally a tale of the field getting interested in learning and transfer, running at it with a basic (and very Cartesian) idea, failing miserably, giving up, and then getting excited again 30 years later, only to repeat the mistakes, just with different names for things. It was kind of amazing to read, and to see the complete lack of awareness that the new idea was literally just the old idea in different clothes. It also makes it abundantly clear that the ecological task dynamical approach Danny is taking is a radically different move, so that if it fails, at least it will fail for new reasons. 

The next lecture is, I think, the last bit of set-up Turvey will do before embarking on an analysis of the history of theories of perception.

No comments:

Post a Comment