Tuesday 9 June 2020

Lecture 4: Simulative, Projective, and Locality Assumptions (Turvey, 2019, Lectures on Perception)

This lecture is a brief history of the common assumptions made in theories of perception about how things 'over there' can cause us to have a given perceptual experience. The simulative and projective elements can be quickly dealt with; the big claim in this lecture is that the right notion of causation for perception is non-local, as it is in quantum mechanics. (Note: Turvey is not saying perception is a quantum process. He's just going to use it as a framing to explain what non-local causation is, and he will rely on the rigorous empirical testing it has passed in physics to say it is a viable notion of causation for a physical system.)

Simulative and Projective Assumptions

Most theories of how the world can cause perceptual experience rely on an assumption of local causation; perceptual experience is the end result of a chain of events in which things cause a change in something they are next to. Causation is not, here, allowed to invoke 'action-at-a-distance'. This relates to the Cartesian, mechanical notions of cause and effect discussed in Lecture 3

In this kind of analysis, one way for something 'over there' to cause a perceptual experience of itself is for the light to carry a simulation/copy; this is the simulative assumption. These copies must then be projected to the observer; this is the projective assumption. Together, these form the basis of indirect theories of perception, and they tend to immediately cause familiar problems. For example, when the geometric properties of light were studied more formally, and principles of projection (e.g. the camera obscura) were considered, it seems to immediately be the case that the simulation being projected is impoverished, and perception then immediately requires some form of knowledge-based enrichment. But where does this knowledge come from, if not perception? Theorists from Plato through to modern times have been struggling to answer these questions. 

Non-Locality (is action-at-a-distance an option?)

Cartesian mechanics denies the possibility of action-at-a-distance; there must be links in the causal chain all the way from one thing to another for the first to cause something in the second. This also implies that causation takes time; cause precedes effect, rather than being simultaneous with it. 'Spooky action at a distance' such as Newton's inverse square law of gravitational attraction eventually got resolved by the concept of gravitational fields, for example, with causation limited to, at most, the speed of light. Collectively, this is the idea of local causation and science tends to look for solutions within this idea.

Quantum mechanics blows this out of the water immediately, which is one reason people like Einstein hated it to begin with. One of the principle features of quantum mechanics is non-local causation, in which one thing is able to cause a change on another thing without any causal chain and with no speed limit. Non-local causes are simultaneous with their effects. The classic example is the uncertainty principle. The most common version we know about is that you cannot know a particle's speed and position simultaneously, but this is just one example of the general principle of non-local causation in quantum mechanics. 

The reason why this is the case is actually something we've already encountered. It's because quantum properties are defined impredicatively, that is, they are defined with respect to the complete system in which they feature. A quantum thing measured by a position meter has a position but nothing else; that same thing measured by a velocity meter has a velocity but nothing else. 

Normally, as we've noted, impredicative properties are things to be avoided; they imply vicious circular causation and the inability to have objective properties of things. Quantum mechanics has actually tried very hard to break it's impredicative properties, but the reason why Turvey is talking about this is that it has completely and utterly failed to get rid of them. Quite the contrary; the evidence that quantum properties are impredicative is as good as it gets in science (Turvey discusses the specific example of the Einstein-Podolsky-Rosen experiment and Bell's Inequality). 

The practical upshot is that impredicative properties and non-local causation are legitimate options for a physical system. This is good, because Turvey then notes that non-local causation is at least implied by the idea of perception as a two-term relation. For example, consider a person visually exploring a room by scanning with their eyes; how does this cause the experience of a constant, unchanging room? Indirect (at least 3 term) theories of perception are allowed to invoke an intermediate step that stitches together the series of views scanning provides. Direct (two term) theories don't have this option, so instead they claim that the act of scanning the room provides access to the specification of the scene. If this is going to work, we're going to need non-local causation.


It was hard to find the nugget of this chapter; the thing it was about. The discussion of quantum physics is pretty long, for example, and the chapter title doesn't actually help. However, I finally realised the core idea was about causation. Two term, direct theories of perception imply non-local causation and impredicative properties as the objects of perception. Turvey spends so long on quantum theory, I think, to ram home the point that both of these are legitimate options for a physical system. The evidence that these are inescapable facts of quantum mechanics is legitimately overwhelming. 

The real question is going to be, what form does impredicativity and non-local causation take in perception and action? It's not going to be exactly as spooky as in quantum theory, but it's still going to be weird, from a local causation point of view. Importantly, I don't think we are going to be able to simple port properties over from quantum mechanics, nor even properties from common sense (such as 'size'. Perceptual properties live at the ecological scale, and we need to consider how impredicativity works there. Ecological psychologists need to put in the hard work of identifying our own properties and testing how they cause perceptual experience (Gibson always emphasised that this is part of our job). 

Relational Affordances

I've also realised this is the core of the objection to my argument that relational affordances cannot be perceived (blog post, preprint). My point is that the relations that get invoked do not typically interact with energy media so as to create information. They tend to be too abstract, such as the relation that means beer cans typically contain beer. The main objection I run into is that I am appealing to a limited notion of cause. It's never been spelled out by these reviewers in a way that makes sense, but I think now they are sitting on this appeal to non-local causation.

My concern has been that they have been relying on it in a way that seems too magical. This chapter makes me realise that while I still think this a little, it seems this way because non-locality seems insane and spooky from a local causation point of view, and I think I have been standing there. So I do need to move, but I remain convinced that information must still be created by something for it to be an abject of perception. Specifically, I think that information forms a critical component of the impredicative mechanism that allows us to measure and therefore perceive things like affordances. Take the EPR experiment; part of understanding the non-locality is understanding the impredicativity, which requires you to specify the entire system the quantum thing is currently a player in (is it part of a position measuring system, or a velocity measuring system?). 

That is a first draft, but I think I know have a toehold for a revision to the paper.

No comments:

Post a Comment