Not Every Behaviour Entails an Affordance

There are two approaches to formalising Gibson's notion of affordances. The first is to treat them as dispositional properties of the environment; properties the environment offers the organism to support possible actions. The second is as relations in the organism-environment system; not localised to the environment but formed when the organism-environment system is formed as a perception-action loop. I am firmly Team Dispositional Properties, because I think this best aligns with Gibson's goals, it allows actual science on the coordination and control of action (Wilson, 2026) and because there is currently no working account of how you might perceive a relational affordance. 

Despite all this, the relational account remains alive, well, and probably the main way people think about affordances. One reason is the drive to make ecological psychology align with the commitments of radical enactivism, a drive I have no interest in (I think enactivism is ok but nothing we owe our allegiance). I think the other big reason is that only the relational account really allows you to talk about affordances for things other than the coordination and control of actions, and as the ecological approach expands the things it is trying to explain people want to apply the ecological tools we have, and that means affordances. 

Except actually, it doesn't. 

Obviously affordances are key; they are central to developing ecological explanations of behaviour that do not entail mental representation or processes like prediction (they are how prospective control works; Turvey, 1992). But Gibson did not build his ecological approach solely out of affordances. Events exist, are real, and can be perceived (Bingham, 2000) and these are not affordances. More importantly, the real meat to the ecological approach Gibson developed is information, and we might be able to learn to use information in all sorts of ways. So Gibson actually provides several resources from which to build ecological explanations, and not every behaviour entails an affordance (one obvious example is Bingham's model of coordinated rhythmic movements, in which affordances do not play a role). 

There's one additional point that's been on my mind. Gibson gave us affordances, events, and information, but he never said these were the only things that exist at the ecological scale that might enable perception, action, and thus cognition and intelligence. His written work was never his final word but just where his thinking was at the time, and the 1979 book was only his last word because he died - he was still continuing to develop his ideas when he passed. 

I think our field is every so slightly stuck in a rut, formed by the assumption that ecologically, behaviour can only be organised with respect to affordances; that affordances are the only way we avoid representations, mental or otherwise. This rut is, I think, the main motivation for defining affordances as relations - if ecologically explaining things like language or other social/cultural behaviours needs affordances then they probably aren't dispositional properties. But I think this rut is a mistake we might be able to get out of other ways, and I am working on the idea that pointing this out is another way I can push back on the relational affordances account that I think is currently fatally flawed (if you cannot perceive them, they cannot be psychologically relevant). 

So. The ecological approach currently contains affordances, but also events and information as resources for building explanations of behaviour. We over-rely on affordances and therefore we may be asking too much of them - they just might not be the kinds of things to support behaviours that aren't the real-time coordination and control of action (what Gibson originally developed them to explain and the main thing we use them for). Other behaviours (e.g. language) might entail the perception of affordances; but maybe they don't, and that might still allow ecological explanations. Maybe languaging is about creating and perceiving the information for speech events, not forming linguistic affordances. Maybe it's not events, but about learning to use information in convention-based ways (vs the law-based ways we typically invoke). Golonka (2015) noted that while all information is specific to some dynamical property, we have to learn to use that information variable as information for that property, and this opens to door to the possibility of learning to use that information variable as information for something else. 

What else might there be? To be honest, I don't know. Affordances, events and information seem to span the options of the ecological space that Gibson laid out. Golonka's varieties of information use may constitute another option (or it might be a viable option covered by Gibson's list). All I want to say here is that there might be other categories of ecological pieces, and if there is information for them (i.e. if they can be perceived) then they might play a role in our cognition. 

The take-away point - not every ecological account of behaviour requires affordances, and we should maybe take more care to ask of each task, what is actually perceived to support this behaviour?