The last two Lectures have laid out the mechanical, Cartesian analysis of the problems of perception. Every flavor of this analysis, right up to an including modern representational accounts, rests on 'loans of intelligence', in which important work that's required to make the system work is simply given to the system. For example, Descartes proposes a set of cognitive axioms to ground the inferential processes required to implement knowledge, and makes these innate (specifically, given by God). This won't do - if we want to develop a naturalised theory of perception, we need an account of where every part of the system comes from.
The Cartesian programme is a form of rationalism, roughly the claim that knowledge is based in reason rather than experience. This chapter reviews the opposing camp of empiricism, roughly the claim that knowledge is based in experience rather than reason. Turvey reviews the approaches of Locke, Berkeley, and Hume, with respect to how they assert a mind can come to know about the world via experience. They end up running into very similar problems as Descartes et al, because of their continued commitment to the mechanistic hypothesis. We are going to have to reject both of these as we progress.
The mechanistic hypothesis is an analysis of the properties matter is allowed to have, and it concludes that matter is
- Inert, passive and changeless
- Organised with respect to mathematical laws
- Understandable by the maths of mechanics
- Always in one clearly specificable, permissible state
- Distinct from mind but related via sensing and reasoning
Turvey has been working to undercut these assumptions with references to more modern analyses of matter and the kinds of properties it is allowed to have (e.g. work in quantum mechanics). What he is working towards is an ecological scale physics that will allow us to reject the mechanistic hypothesis as the correct basis for an analysis of perception and action. The empiricists try to explain how a mechanical, material system can come to know about it's world via sensory experience (rather than given knowledge) as the primary means. - Locke tries to build it from a realist point of view, where the system works on the assumption that the world is real and causing it's experiences. This means you might be able to use structure in experience to get to structure in the world. (Turvey will note that there is a loan of intelligence lurking here - Locke appeals to God regularly, for example).
- Berkeley takes an idealist stance, where all you have is experience and you have no basis to think that structure in that experience is being caused by any kind of structure in the world. This tries to dodge the loan, but ends up casting the person adrift.
- Hume is the skeptic - he thinks that there are no kind of causal relations in experience, but that the experience sure is organised as if there are and we can therefore act as if there are too. (Michotte eventually does experiments showing that we can, in fact, perceive causation, which feeds into Gibson's thinking very explicitly. I also blogged about this topic here).
There are loans of intelligence lurking throughout these accounts. This shows up most clearly in the criteria these thinkers apply to experience in order to use it as the basis for knowledge of the world. For example, they appeal to various forms of correspondence (experience of the world can be true it that experience corresponds to something in the world) and coherence (if experience is well behaved, it must be related to something well behaved). None of these work in and of themselves - they all require external support (e.g. someone to evaluate the correspondence) and this support requires a loan of intelligence. You can also have a pragmatic notion of truth (if it works, it's true) but this requires action in order to generate consequences, and so far no-one has action in their theories of perception. (I'll note here that part of the argument Hoffman builds for his interface theory is to mathematically show you can have a coherent experience set that has no relation whatsoever to a coherent world set. He settles on a pragmatic account, his Perception-Action-Decision loop. The funny thing about his critique of representational accounts is that it is bang on; he just makes the mistake of accepting this weirdness is all we have to work with). Hume, recognising these issues, appeals primarily to a 'mental mechanics' that operates in a similar fashion to physical mechanics. But he's committed to the mechanistic notion of physics, and this account fails to provide any room for meaning (intentionality) in our systems. A Glimpse at the Solution
Turvey now completes his introduction to the basic ideas from modern physics we are going to need. He discussed non-locality and the implications it has for causation in Lecture 4. The other aspect is self-organising physics - the analysis of physical phenomena that are anything but passive and inert. These are (eventually) going to let us abandon the mechanistic hypothesis and it's problems, and embark on a genuinely different way of understanding perception. He discusses the example of the Belousov-Zhabotinsky reaction, in which the chemistry of a mixture of some organic and inorganic components leads to a solution that oscillates between two states. The solution also never settles into either state, because the underlying processes are never able to reach balance. These now common examples are physical systems that are organising themselves, in direct contraction of the mechanistic hypothesis. A final consequence of the new physics is the possibility of a general commitment to the notion of complimentarity (in which two formally incompatible descriptive modes are both required for a full accounting of a physical system). For example, light is neither a particle nor a wave; which properties it exhibits depends on how we measure it and both descriptive modes are required to understand how light works. Complimentarity is the correct level at which to analyze the system. Reflections
This chapter marks the end of the preliminaries. Turvey has been laying a lot of groundwork, making some promises, alluding to how those promises will get fulfilled, and generally giving us a vocabulary to take with us as we move into the specifics of perception. Honestly, it's been a slog to pull these out, and Turvey's writing style has been a pain in the ass - he does very little of the work of drawing these connections. I think this may be a side effect of this being the written material for a course that is very heavy on discussion, which I get but I still think is a problem I am trying to solve with these blog posts. I was very interested when I realised that Turvey's analysis of empiricist attempts to relate experience to the world tracks Hoffman's analysis of the problem with his interface theory. This is something I struggled with as I critiqued Hoffman; he's right about a lot of things! I finally realised that Hoffman's analysis of the nature of representational theories (which, by Turvey, are generally Cartesian and mechanistic, and occasionally Humean and mechanistic) is absolutely correct. He effectively goes over all the ground Turvey has worked through up until now, but instead of noticing that there are unpayable loans of intelligence everywhere, Hoffman doubles down on the problem and claims that it is a feature, not a bug - interfaces are weird but just the ways things are. It backs up my analysis that his critique of Gibson fails because the ecological approach is an entirely different kind of analysis of the problem of perception. Hoffman's paper would make a good addition to the reading list for this course.