Monday, 10 May 2010

Affordances, Part 1: Affordances are real dispositions of the environment

It turns out I've been pretty confused about affordances for a long time. This is partly due to the fact that I don't (yet) do research on affordances and so don't spend much time thinking about them. However, it is also partly due to the fact that affordances are fucking weird and much has been written that is confused, incomplete and wrong.

I got back into thinking about affordances in detail by reading Anthony Chemero's mostly excellent new book, Radical Embodied Cognitive Science (Google Books preview). Actually, everything I've written so far is actually me getting my head in the game to go after something I think Tony has gotten needlessly wrong; for various reasons he thinks its time to lower the specification requirement for perceptual information. I think the reasons are flawed and that it's not even close to being an empirical reality that specification is not needed; but this is all material for later on.

When Tony talks about affordances, he proposes that they are relations. He contrasts this to the generally accepted Turvey formulation, that affordances are properties, specifically dispositional properties. This, it turns out, is a key internal battle that is ongoing within ecological psychology. I was initially on Tony's side: Turvey's account has always seemed wrong to me. But after some discussion with my PhD advisor, and after reading Heft's summary of what Gibson meant, I'm back on the side of affordances as properties.

So, Affordances, Part 1: Affordances as Dispositional Properties of the Environment (Turvey, 1992)

Turvey (1992) lays out an ontology for affordances that allows prospective control (i.e. 'planning' actions). Prospective control is about realizing an action goal; you therefore need to know about the possibilities for realizing your action goals; thus affordances, which are about the possibilities for action. Affordances must be perceivable so they can therefore serve to allow prospective control of action.

Laws ("invariant relations between or among properties of things") prescribe what can possibly occur; for something to actually occur, the circumstances under which the law obtains must come to pass:
Laws and circumstances (auxiliary conditions, boundary conditions, initial conditions) yield actual state of affairs. (Turvey, 1992, pp 177)
The ecological research programme, of identifying possibilities for action and how they come to actually occur, is therefore a search for laws and conditions. Affordances are the law part of things (i.e. what can possibly occur) - so what kind of properties are affordances?

For Turvey, affordances are dispositional properties of the environment. Dispositions are odd properties: the classic example is the solubility of salt, which is certainly a property of salt but one that is only apparent when salt is placed in water. Dispositions have three primary characteristics (pp 178):
  1. A disposition to do Y is prior to actually doing Y. Salt was soluble prior to ever having been exposed to water
  2. Dispositions come in pairs: salt is soluble and water can act as a solvent. There are two complementary parts to a disposition, which Turvey will call the disposition and the effectivity. Which is labelled as which depends entirely on your focus; water's propensity to act as a solvent to certain materials can be a disposition, or the effectivity of the disposition of salt to dissolve. (This actually matches nicely to Gibson's descriptions of affordances and the Jamesian notion of selection as a key component for experience)
  3. Disposition + effectivity  = actuality. When the entire system is present, the disposition is always effected/actualised.
The logical structure of this is derived from Bunge (1977). That said, it goes as follows.

Let X be some feature of the environment (e.g. a surface with the properties of an adult’s chair), and Y be an organism in that environment (e.g. an adult human). Let p be some property of X (sit-on-ability), and q be some property of Y (appropriate size). Then p is an affordance of X and q is the effectivity of Y if and only if there is a third property r, such that
  1. The system comprising (X+Y) possesses r;
  2. The system comprising (X+Y) does not possess p or q;
  3. Neither X nor Y by themselves possess r.
r is therefore a relational property of a higher order organism-environment system (salt-dissolved-in-water; adult-human-seated-on-chair); but crucially the affordance, p, is a property of the environment, and not the relation. The affordance can thus structure ambient light, and be perceived according to the rules of ecological optics.

Step 2: This, I admit, I still do not understand. Turvey describes this condition by noting that the solubility of water (p) comes from the nature of the electrical attraction between ions, which no longer exists in dissolved salt. This I understand: how this relates to action I do not yet know; the adult human still has the same dimensions, for example (although it's possible that this is simply the incorrect way to describe the key properties); ideas on this welcome.

A final note: the (Turvey) ecological research programme is to seek laws containing affordances, and establish both the details of the law and the conditions under which the law obtains. If your data don't suggest a lawful relationship, you have yet to run the correct experiment and should keep pushing. This is the practical consequence of the Turvey ontology. I flag this up because Tony's book objects to this practical consequence and therefore needs to object to the ontology, which I think is an error. He is, at least in good company and clearly inspired by his reading of Heft on James.

Bunge, M. (1977). Treatise on basic philosophy (Vol. 3). Boston: D. Reidel.

Chemero, A. (2009) Radical Embodied Cognitive Science. Cambridge, MA: MIT Press.,, MIT Press e-book

Turvey, M. (1992). Affordances and Prospective Control: An Outline of the Ontology Ecological Psychology, 4 (3), 173-187 DOI: 10.1207/s15326969eco0403_3

No comments:

Post a Comment