Chemero (2009) Chapter 7: Affordances, etc (Pt 1)

If you want perception to be direct (no 'mental gymnastics') you must identify where the content of perceptual experience comes from; when I view a chair, for example, I don't see a meaningless or random collection of surfaces or colours, I see an object that I can interact with in some ways and not others. For traditional, indirect theories of perception, this meaning is constructed internally: mental representations perform transformations (perhaps computational ones) on sensory input to infer what the input means. A theory of direct perception requires that meaning is not added to the signal; this 'enrichment' is not permitted. To solve this problem, Gibson proposed that the world, for a perceiving-acting organism, is not comprised of meaningless bits of physics and chemistry. Instead, he proposed that the world presents itself to this organism in terms of affordances, which are intrinsically meaningful for the organism and are about the possibility of behaviour.

So a theory of direct perception requires an ontology, a theory about the make-up of the world  that means it is intrinsically meaningful. Chemero wants RECS to include direct perception, therefore he needs such an ontology. He is happy with affordances; he is not, however, satisfied with the Turvey-Shaw-Mace approach which defines affordances as dispositions. This chapter will defend an extended version of his theory of affordances as relations: Affordances 2.0.

Affordances have cropped up here, er, a couple of times already - namely here, here, here, and most recently and enthusiastically here, as well as some empirical discussions here and here. The idea that affordances, not physics, is the correct ontology for a theory of direct perception also came up here. To jump ahead, I disagree with Chemero about affordances. I don't think the disposition account is flawed in the way he thinks, and I don't think relations solve the problem anyway. While I obviously agree that an affordance-based ontology is the way to go, I am so far satisfied that the dispositional account is the best current analysis, and I think that trying to make affordances relational is to confuse the world with information about the world.

I'm going to take this chapter in two parts, because this post got long; affordances are complicated things. After this chapter, I think a pause for station identification may also be in order, just to lay a few things out in response to Chapters 6 and 7, the real meat of the book. But first, Affordances 1.0 vs Affordances 1.1.

Affordances 1.0
The first attempt to formalise Gibson's definition was, of course, Turvey et al (1981; Turvey, 1992). This defines affordances as dispositions, properties of the world which are complemented by the effectivities of an organism. Framing affordances this way allows them to be defined with reference to the organism; the complete description of a disposition requires the conditions that allow that disposition to manifest itself, and for affordances that includes the organism.

Chemero then distinguishes a second, related version of affordances, in which they remain properties of the world but are not dispositions; instead, they are resources of the environment, properties of objects which an animal might use (Reed, 1996). This makes them a source of evolutionary selective pressure, because they exist prior to the presence of an organism that might come to use them.

Reed's account is selectionist, and tied to evolutionary biology. Chemero contrasts this to the more physics-oriented approach Turvey-Shaw-Mace (TSM) approach. Dispositions can't exert selection pressure, because properties of the world aren't dispositions unless complemented, so  properties of the world aren't affordances (by TSM) until an organism shows up. A key question which follows this is to ask what aspects of the organism complement the affordance; TSM suggest effectivities, other empirical work has suggested body scale (e.g. defining the affordance for stair climbing relative to leg length; Warren, 1984).

The compulsory nature of dispositions
This comes up later in the chapter (p. 145) but is relevant now. Part of Chemero's critique of affordances-as-dispositions is that when the disposition and it's complement are present, the disposition simply manifests - it is compulsory and guaranteed. Dispositions do not fail to manifest when all the conditions are present. Chemero thinks that abilities like 'being able to walk' can't be dispositional (effectivities are  dispositions too, just mirrors of affordances) because they can fail even when all the necessary conditions are met. Abilities must be functions of an organism's personal and evolutionary history, thus allowing for malfunctions.

Some comments
1. Dispositions
Dispositions do manifest when the conditions are met; but for a complex disposition such as an affordance or effectivity, it seems unproblematic that the required conditions are simply not always met. For instance, I am currently seated - I am literally currently not a walking device, and thus I am in no danger of trying to effect the disposition of the floor to support my locomotion. I can become such a device, and if I manage this I will indeed manifest walking behaviour - but there are many reason why I might fail to establish all the necessary conditions. But critically, I am not always capable of effecting every affordance in my environment. I am a task-specific device (Bingham, 1988) and must become a different one to do anything different.

2. Do affordances persist?
Chemero summarises Affordances 1.0 as describing affordances as properties of the world, with some disagreement between TSM and Reed about whether they exist independently of animals and precisely what properties of the animal they should be defined with respect to. The persistence issue comes up again below, because Chemero thinks only relations solve the problem.

To be honest, I'm not sure I see the disagreement and I don't think there's a problem for the dispositional account. My arguments with Ken Aizawa about affordances and the TSM approach pointed me to the key discussion of anchoring properties in Turvey et al (1981). These are the physical properties of the object which underpin the disposition; salt is soluble in water because of the electrical properties of the ions in the salt. These properties persist. It seems that there is a unbroken chain of links from the world to the dispositional property, such that the entire system is firmly grounded. The physical properties of objects predate organisms, and make it so that a given object non-accidentally has certain characteristics. These act as Reed's resources, defining a niche which can exert selection pressure on an organism (although there are apparently concerns about being selectionist which Chemero mentions but does not go into). Organisms then come to complement the anchored dispositions, and once this process is up and running the dispositions are capable of continuing to exert selective pressure.

This argument needs work, clearly: but as a first pass it seems that there is a perfectly sensible story to be told in which affordances are real properties of the world, if you allow yourself the full range of tools (dispositions and the idea of anchoring properties).

Affordances 1.1
Moving on, Chemero now lays out the basics of his original relational scheme. First, he motivates the move with two arguments.

1. Affordances entail feature placing
Chemero (2001) proposed that a solution to these controversies lies in rejecting the idea that affordances are properties of the world, or even properties at all. Instead, he suggested affordances are relations, between particular aspects of animals and particular aspects of situations (remember, Chemero has broadened his informational account to that covered by situation semantics). 

He first notes that affordances seem to be an example of feature placing (Strawson, 1959). The example is the difference between noting your car is dented (identifying a property of the world) and noting that it is raining (placing a feature). What we perceive when we perceive affordances, Chemero argues, is more like noting that a situation has some feature (e.g. perceiving that 'it's time to flex your elbow'; Michaels, 2000). Features are not properties, and therefore affordances aren't either. They also aren't features of the world, they are features of entire situations (which includes the organism) - affordances therefore span the world and organism. Specifically, for Chemero, they are relations between these parts of the situation.

2. The problem of two minds
His second point relates to a topic that came up in Heft (2001) and which I blogged about previously: the problem of two minds. The issue is that if two people are directly perceiving the same thing, that thing forms part of the perceptual experience for both people; their minds therefore overlap at the object. Chemero notes affordances, if they are directly perceived properties of the world, suffer from the same problem. You can solve this problem, as Heft noted, if you assume that each person is in a unique relation to that property; I am seeing the object from one place, you are viewing it from another. The relation doesn't overlap, and thus there is no problem.

A comment: Chemero takes this as support that the affordance must be a relation. But as I read Heft, he actually (correctly) places the Gibsonian solution in the information, and not the world. Heft notes that the relational aspect comes naturally from the ecological analysis of the optics - the object is effectively 'multiply realised' in the optics, because each point of observation has a unique view of the object. The act of perception is relational; what you are perceiving is not (I will suggest).

Affordances 1.1
Gibson's ecological psychology is a form of realism, specifically one in which relations are perfectly real things; this means that affordances being relations doesn't immediately cause any ontological trouble for RECS. Relations define the way in which two things (the relata) are connected: for instance

   taller-than (Shaquille, Henry)  (1)

means that Shaquille is taller than Henry. For affordances, the relata must include the environment and the organism:

   Affords-behaviour(environment, organism)  (2)

The environmental relata must be features, not properties (p. 142). The organismal relata might be things like body scale, but could more usefully be described as an ability of an organism. Bill Warren ran a key study on affordances, in which he asked people to rate whether they could climb a set of stairs with a given riser height; different people switched from 'yes' to 'no' at different riser heights, but these individual differences vanished when the data were expressed as a function of the individual's leg length (as pi numbers; Warren, 1984). The suggestion is that people perceived the stairs, not in absolute terms, but relative to their own ability to step up. Leg length (body scale) is an easily measured proxy for ability, specifically the ability to climb the stair, an idea backed up by some empirical work by Chemero. So (2) becomes

   Affords-behaviour(feature, ability)  (3)

Perception of this affordance is, itself, a relation:

   Perceives[animal, affords-behaviour(feature, ability)]  (4)

but our phenomenal experience is not of the constituent relata, just the affordance:

   Perceives[animal, affordance-for-behaviour]  (5)

So affordances are relations; the perception of affordances is a relational act; as a flavour of realism, relations are real and so, therefore, are affordances and the perception of them. Everything RECS needs exists - so far so good.

Three brief points then follow, questions supposedly answered by Affordances 1.1 (affordances as relations):

1. Different animals may exist in the same place but occupy different niches, which Gibson took to be the set of affordances for an animal. Affordances 1.1 solves this supposed problem in the same way as the problem of two minds - different animals have different abilities, and thus stand in different relation to the same features of the environment. Like the problem of two minds, I'm not sure there's much of a problem here and if there is, it's solved in information, not the world.

2. Stoffregan (2000) was worried that if events and affordances were different, there would be no reason for an ecological psychologist to think events were perceivable. For Affordances 1.1, events are perceivable because they are a change in the relation between an organism's abilities and features of the environment, i.e. a change in the layout of affordances.

3. If affordances are dispositions, what happens to them when the complementary organism isn't present? Chemero thinks this is a problem for a supposedly realist ecological psychology; his solution at this point is to simply claim that affordances are like being 'lovely' (Dennett, 1998):

Dennett distinguishes between things that are lovely and things that are suspect. To see the distinction, consider that a female hippopotamus in a zoo might be lovely, even if no male hippopotamus has ever seen her. She is lovely just in case if a male hippopotamus were to see her, he would find her to be so. The key is that being lovely depends on a potential observer, not an actual act of observation. Compare this to being suspect. To be suspect, something actually has to be under suspicion. Being suspect requires an actual observer. Whether affordances exist without animals is a matter of whether affordances are lovely or suspect. Affordances, we can see, are lovely. A feature of some situation might exist just as it is even if there are no animals. There will be affordances in which that feature takes part as long as some animal exists with the appropriate ability.

Chemero, 2009, pg. 149

This analysis holds for relations, but it's not immediately apparent that it's not just as true for dispositions. In fact, it sounds exactly like dispositions, especially if you remember about anchoring properties.

Some thoughts
The next section is on Affordances 2.0, an expansion of the affordances-as-relations accounts. But so far I'm not convinced;

1. I think there is a confusion lurking here. Gibson worked very hard to keep the world and information about the world distinct. These things are not identical; as a simple example, the information for relative phase is not relative phase, but the relative direction of motion. Chemero is correct to note that relations solve the problems he notes: I just think he has put the relation in the world, instead of in the optics, and that this is an error. The critical relation is the act of perception, and what is perceived are properties. 
2. I have to thank Ken's assault on the TSM approach for making me aware of anchoring properties; it's been a long time since I read the laws paper in that much detail. I think these serve the critical role of grounding the affordance to the physical world.
3. I also have the hunch that the account of dispositions currently available to ecological psychology is out of date; since Turvey (1992) there's been a major research programme in philosophy about dispositions, which I believe was kicked off by Stephan Mumford (1998). I would actually like to bring these developments to ecological psychology, because I think we might find it useful. Maybe I'll send him an email one day. Regardless, there's work available to improve any problems with the dispositional account before we need to throw it out. 

References

Dennett, D. (1998). Brainchildren. Cambridge, Mass.: MIT Press.

Heft, H. (2001). Ecological Psychology in Context: James Gibson, Roger Barker and the Legacy of William James's Radical Empiricism. Mahwah, NJ: Lawrence Erlbaum.

Michaels, C. F. (2000). Information, perception, and action: What should ecological psychologists learn from Milner and Goodale (1995)? Ecological Psychology, 12 (3), 241-258.

Mumford, S. (1998). Dispositions. Clarendon Press.

Reed, E. (1996). Encountering the World. New York: Oxford University Press.

Stoffregen, T. (2000). Affordances and events. Ecological Psychology, 12, 1-28.

Strawson, P. F. (1959). Individuals. London: Methuen.

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


Warren, W. H. (1984). Perceiving affordances: Visual guidance of stair climbing. Journal of Experimental Psychology: Human Perception and Performance, 10, 683-703.

Tweet