Affordances something to think about?

A debate is/was going on the proper use of affordances. Often it has been misused, and some claim, because of that, affordance needs to be redefined/refined. This essay argues that affordance as a defenition is not at all ill-defined, but people should simply stop misusing it: Cognitive affordances are a misnomer.

In 1999 Norman wrote an article pleading for a more deliberate use of the term 'affordance'. According to him this term has often been misused in Human Computer Interaction (HCI) literature and discussions. Especially when considering the typical Graphical User Interface (GUI) interaction using keyboard, mouse and screen. He would regret it when the term affordances was to be dropped for being too ill defined. Several years later, Hartson (2003), in agreement with Normans plead, replies to Norman proposing four types of affordances. 

In this commentary I will first introduce Hartson's (2003) proposal, compare it with the use of the term affordance in other domains and finally attempt to judge whether the distinction would be a valuable contribution to HCI-research and design and/or other domains. 

Physical affordance, or real affordance (Norman, 1999), is referring to the original meaning of affordance once introduced by Gibson (Hartson, 2003), but neither Norman, nor Hartson relate their definition to the original definition by Gibson. Therefore, in short, what some of my textbooks say about Gibson's affordance: “affordances are something that objects manifest, specific activities that are afforded by an object” (pp. 236-237, Sekuler & Blake, 2002). Typical examples are a chair affording sitting, and a cup affording drinking. But a cup can also afford scooping and a chair may afford, by standing on it, grasping higher placed objects. Affordances allow us to perceive objects for their ecological relevant functions and relates to perceiving how one can interact with the environment (Bell, Greene, Fisher & Baum, 2001). In HCI this is a design feature that helps users in doing a physical action in the interface, but often HCI literature has used the term affordance to refer to what will from now on be referred to as 'cognitive affordance' (Hartson, 2003; Norman, 1999). 

Cognitive affordance, or perceived affordance (Norman, 1999), are to Hartson (2003) design features that help users in knowing something. Norman's definition is about what a user perceives to be possible. An example of such an affordance is for example the button label. Even though Hartson claims his Cognitive affordance is similar to Norman's perceived affordance, they differ on a crucial aspect: the physicality of virtual objects. Norman does not want to call buttons on a display real affordances: “Sure, you can click on the object, but you can click anywhere” (p. 40), expressing his need for physical differentiation between objects. Yet Harson is “treating active interface objects on the screen, for example, as real physical objects” (p. 319). Since Hartson distinction is the most precise, I will continue to work with his definitions. 

Sensory affordance is implied in Norman's paper (1999), but Hartson (2003) believes there is a need for emphasizing on the perceptibility of the object. It is a design feature that “helps, aids, supports, facilitates, or enables the user in sensing” (p. 322) and can be considered an attribute of a cognitive or physical affordance. Making a label larger, for example, increases sensory affordance. 

Functional affordance, to end the list of affordance types as defined by Hartson (2003), is defined as the type of functionality that is offered by an object or system (the latter referring to the higher level applicability of the definition). It helps or aids the user in doing something. Hartson proposes to always include a purpose in the definition of a physical affordance to “bring Gibson's ecological view into contextualized HCI design” (p. 321). Considering function is another way of discriminating between clicking on functionless white space compared to clicking on a button with a function attached. A difference emphasizing on the physicality of objects with a function. Here Norman and Hartson are in conflict: recall that Norman (1999) rejected the existence of physical affordances in the Graphical User Interface (GUI). 

Is 'affordance' indeed so ill-defined? HCI is just one of the domains in which the term is being used. How do other researchers use the term affordance? Better understanding of the use in other domains may give a fresh insight in the use of affordances and might enable judging whether these domains could learn from the insights by Hartson (2003). The articles now discussed were part of the Advanced Cognition course this assignment is part of. 

First I will shed a closer look on the use of the term affordance by tangible interface explorers Ishii and Ullmer (1997). In their introduction they show their admiration for the aesthetics and rich affordances of historical scientific instruments. The continue saying: “Through grasping and manipulating these instruments, users of the past must have developed rich languages and cultures which valued haptic interaction with real physical objects.” Affordances not only add functional use, but also add to cultural wealth in terms of expression and vocabulary. Ishii and Ullmer continue with stating their intention to make better advantage of the natural physical affordances, referring to Norman's Psychology of Everyday Things. In a later work (2001) they clarify their interpretation of physical affordances: affordances “also speak to people’s ability to creatively combine physical objects in unexpected fashions” (p. 12). They believe this is hardly possible with only digital artefacts using the traditional GUI-style interaction, but they do not explicitly reject the existence of physical affordances. 

Isshii and Ullmer (1997) do however highlight another issue: they talk about affordances as things that enable creatively combining physical objects. The original concept of affordances also seems to allow for creative use of objects around us. We do not only grab a chair for using it as something to sit upon, but also in order to get a book from the top shelf. We use the environment to get things done.

Haselager's (2005) concept of the embeddedness part of embedded, embodied cognition (EEC) closely connects to affordances. His notion of embeddedness is similar to that of Gibson's ecological perspective: behaviour takes place in the environment. “It is not so much where you are, but what you can do in a situation” (p. 16). One of his examples demonstrating the importance of the environment is interestingly enough an observation of Tetris-players by Kirsh and Maglio (1994, as cited in Haselager, 2005). Tetris is a game based upon a GUI style interaction and yet it allows for exploring through manipulation also being referred to as scaffolding (Clark, 2003; Haselager, 2005). It allows for 'thinking' outside the human mind, manipulating the outside world to make getting things done easier.

I can confirm Hartson's (2003) motivation for writing his paper: the use of affordance varies greatly. First of all, some talk about affordances as an object in and of itself, while others talk about affordances as properties, or manifestations, of objects.  Secondly some, notably Norman (1999) and Hartson (2003), seem to disagree whether physical affordances can exist in a GUI. And finally there is the ability of affording scaffolding neglected in for example the articles by Norman (1999) and Hartson (2003), which is so much emphasized upon by Ishii and Ullmer (1997) and Haselager (2005) when they talk about affordances.

Since so much literature is available that may have spoiled my concept of what affordances are, it is difficult to get a definite answer to the question what would be good use of the term affordance in the HCI-context. In the final part of this commentary a discussion will follow on the issues regarding to affordance use earlier noted. 

The first disagreement noted was between Norman (1999) and Hartson (2003). Norman does not think physical affordances could exist in a GUI environment while Hartson does. Hartson argued a GUI-button could afford clicking and for a reason: it can be clicked upon to fulfill a functional affordance. Thus according to Hartson, affordances should have a function. But isn't an affordance, according to Gibson, perceived by the cognitive system the functionality itself afforded by an object to a certain cognitive system? Despite this comment on Hartson's (2003) argument for the existence of physical affordances in GUIs, I support his statement: physical affordances can exist in GUIs. Take for example the concept of windows. A window affords holding an application. It can be used to move an application around, minimize it, etc. But besides such functionality, it also affords hiding information from windows below it, it can be used as a place holder in case of a 'tile windows' operation, etc.. An object inside the GUI-environment may be less flexible, and thus be limited in the affordances it may offer, but it can at least afford the function it has been designed for.

An argument that might be underlying Norman's (1999) rejection of the existence of physical affordances in the GUI-environment is that the experience is mediated, which could limit flexibility. An optimistic counter argument would be along Clark's (2003) view of hierarchically organized components that can even be extended beyond our bodies. The neural communication channel is only partly replaced by a visual/haptic channel. 

Flexibility is a property of most affording objects emphasized upon by Ishii and Ullmer (1997) and Haselager (2005). This is neglected in all the definitions by Hartson (2003), and may be also often neglected in HCI literature talking about affordances in general. In my view, the flexibility discussed is not really about affordances itself but simply the result of multiple affordances offered by an object. Physical objects may expose a much richer set of affordances, but that doesn't imply objects in a GUI-environment may not have any affordances at all. 

This discussion leaves us with the question whether cognitive affordances are really affordances, and whether perceptibility and functionality are separate affordances? If we want to make a distinction between (technically) mediated and not (technically) mediated affordances, a mediation that might consume cognitive effort, the distinction has some value (but then why not use the term mediated affordances instead?). 'Cognitive affordance' however, seems like a contradictio in terminis. The concept of affordances is based around a direct link between perception and action, bypassing conscious processing, while cognitive affordances demand conscious processing. Affordances were defined as something that is directly perceivable functionality depending on the situation and the perceiver. Introducing 'cognitive affordances' will probably only contribute to the confusion to the concept of affordances in HCI and in general. I would also reject the use of the terms sensory and functional affordances. These are not affordances in and of themselves, but are a necessity for something to be called an affordance or perceived functionality.

I do not comment here on the framework that is being discussed by Hartson (2003) in the second half of his paper. Despite of my comments on the terminology introduced, I acknowledge the need for solid terminology that contributes to locating and in the end solving the right problem. I just would not call everything that has to do with functionality and interaction affordances, that would waste the value of affordance in discussion. 

I believe affordances are pretty much well-defined outside HCI literature and HCI researchers should simply be more careful when calling something an affordance. Adding an adjective does not change that: interaction designers should be taught and use the meaning of affordances as intended. Wasting the very interesting idea of affordances due to misuse would be regretful indeed. 

Bell, P.A., Greene, T.C., Fisher, J.D. & Baum, A. (2001). Environmental Psychology (5th ed.). Belmont, CA: Wadsworth.

Clark, A. (2003). Natural-Born Cyborgs. Oxford: Oxford university Press

Hartson, H. R. (2003). Cognitive, physical, sensory and functional affordances in interaction design. Behaviour & Information Technology, vol 22. no. 5. pp. 315-338.

Haselager, P. (2005). The 7 representational headaches of Cognitive Science. Draft, 6th version, in progress

Ishii, H. & Ullmer, B. (1997). Tangible Bits: Towards Seamless Interfaces between People, Bits and Atoms. Published in the Proceedings of CHI, March, pp. 22-27.

Ullmer, B. & Ishii, H. (2001). Emerging Frameworks for Tangible User Interfaces. [final draft] Is published in John M. Carroll (Ed.). Human-Computer Interaction in the New Millenium (pp. 579-601). Boston, MA: Addison-Wesley

Norman, D. A. (1999). Affordance, Conventions and Design. Interactions, May, 1999, pp. 38-42

Sekuler, R., Blake, R. (2002). Perception (4th Ed.). New York: McGraw-Hill.