Affordances as the innate analogical capacity of a PIF

An ‘affordance’ is something that I can employ in an environmental feedback function for controlling a given perception. The term is often reserved for unusual such pairings of purpose and means. Yes, I can talk of a chair as an affordance for sitting, but discussion of affordances usually goes to sitting on something like a stump, log, barrel, or retaining wall that happens to be nearby.

Don Norman, in The design of everyday things (a wonderful book that John Kirkland mentioned in passing at the 2022 IAPCT conference) gives an example of a switch panel that he had designed to control the lighting in his house. Rather than a row of switches, the panel was a schematic map of the room, with the location of each light corresponding to the location of its switch. But in his first implementation as a horizontal panel, it unfortunately satisfied the PIF for visitors looking for a place to set down their drinks, with unfortunate consequences, so he rebuilt it as a slanting surface so that it would no longer be an affordance (he used the term) for that purpose.

Ignoring the word bridges to nowhere that Gibson has built from this verbal bridge abutment (B:CP p. 167), we can still talk about how we recognize that something unusual can serve a usual purpose.

Recognizing an affordance in the absence of a customary perceptual input in the Sequence perception ‘sitting down’ (a piece of furniture designed for the purpose) is just a matter of perceptual input matching the PIF for the final step of the sequence ‘sit down’. A flat surface, well supported at about knee height, is perhaps the relevant configuration perception that is part in common of the configurations that we call chair, stool, bench, couch, etc.

Bill has an interesting discussion of Gibson here.

1 Like

An affordance is of no value in perceptual control if you don’t know how to use it (have not reorganized to include it in control of some perception). That’s basically why Kent and I invented the word “atenfel”. An atenfel necessarily incorporates an affordance but also necessarily includes the ability to use it in control of some specific perception.

An affordance just sits there in the environment forever, regardless of whether anyone ever learns to use it for some purpose (control of some perception). The word “affordance” has no usefulness in perceptual control theory, which is about the control of perceptions, and the implications of controlling perceptions. Anything that makes sense in that context does so as part of a functioning perceptual control loop.

Of course, you can’t complete a control loop through an environmental feedback path without having suitable affordances somewhere in the loop, just as you can’t complete a control loop without suitable sensors to sense the environmental change that is perceived as change in a controlled perception.

This is interesting discussion. I have heavily utilized and somewhat developed the concept of “competence” (especially in the context of education, e.g. my article in the Interdisciplinary Handbook II which should quite soon be published). The relationship between these three concepts is like this:

Atenfel = Affordance + Competence

If one or another of the partial concepts, affordance or competence, is missing then there is no atenfel and control of perception is impossible. Without the the corresponding other side affordances and competences are just something virtual waiting for their realization as an atenfel.

So if you want to design affordances into our environment, you must start from respective competences. In another way around if you want to educate some competences to control some perceptions, you must take into account what affordances there could be in the environment by using of which the control could take place.

Hi Bruce,

It’s funny to follow how Norman changed his mind about affordances. His “action theory” appeared in full length in User Centered Design, the same chapter where I discovered B:CP. Here’s the footnote I mentioned in my presentation (p.38):

He didn’t mention B:CP in The design of everyday things, even though that text is an expanded version of the same ideas. Footnotes for chapter 1 (p.221) are very similar:

And before that note, he stated his divergence with Gibsonian affordances (p.210):

Anyway, most interaction designers I know assume Norman’s “perceived” affordances are the same developed by Gibson. In the revised and expanded edition (2013) of his book, Norman proposes a new concept called “signifiers” to try solving the mess:

I think the best part of his previous work, which resembles PCT in more interesting ways, concerns action slips. He discusses how a hierarchy of feedback systems would counteract unintended errors during sequences of actions. No references to Powers or PCT, though:


The way I see the whole history, from a interaction design perspective, Eetu’s semiotic, value, and behavioral competences could help building a PCT explanation to Gibsonian affordances, without commiting oneself to direct perception. For instance, Tony Chemero (a modern Gibsonian), in Radical Embodied Cognitive Science, argued that animals are “attuned” to niche-specific information in their environments.

I didn’t know Martin’s and Kent’s “atenfel”, so I’ll check it out. Thanks for the tip!


We would agree that, because the organism is ‘coupled with’ the environment, there is no need to posit within the organism a symbolic representation and computations thereon, and for everything that such notions aim to explain control of perceptual input is sufficient and necessary explanation. The organism acts as if internally-constructed perceptions are in the environment, but (in a stock example) the taste of lemonade is not in the environment. Arguably ‘information in the environment’ is the basis for our constructing that taste perception, but only if we reify perceptions constructed in chemistry and physics, and even then that is at a scale and granularity so far removed from the taste perception that we can’t help noticing that the notion of information in the environment is pretty near vacuous. (I think Vlatko Vedral would say completely vacuous.)

To communicate PCT and help people reorganize their understandings we have to disturb some of their controlled variables. It behooves us to do so gently, so their control systems do not perceive an attack and control to avoid us, and it is essential that we first assume the point of view of the organism. It is only from within their current conceptual convictions that a way to PCT can become evident to them. If with temporary willing suspension of disbelief we join them arguendo, recognizing contradictions and other problems from their point of view can open a way to PCT. For example, the popularity of Norman’s work provides a common basis on which this can happen. Just saying oh he’s wrong here, and here, and there raises the bar and conflicts with the purpose of teaching and promoting PCT.

1 Like

I totally agree. Don Norman is a fantastic communicator and did a great job translating scientific findings to non-academic designers, engineers, and technicians, using an accessible practical approach. I don’t know if the comparison is fair, but Tim Carey’s “Control in the Classroom” does a similar great job promoting PCT to teachers. All the important bits are there, but their presentation is gentle and inviting.