E–Learning, Volume 2, Number 4, 2005 doi: 10.2304/elea.2005.2.4.

402

The Problem with Affordance

MARTIN OLIVER
London Knowledge Lab, Institute of Education,
University of London, United Kingdom

ABSTRACT This article reviews the concept of ‘affordance’, a term widely used in the literature on
learning and technology to try and explain the properties technologies have. It is argued that the
concept has drifted so far from its origins that it is now too ambiguous to be analytically valuable. In
addition, it is suggested that its origins in ecological perception – where relatively primitive claims are
made at the level of species about animal-object relationships, framed in an evolutionary timescale –
have little direct relevance to the moment-to-moment interactions of an individual using a specific
artefact. Because of this, it is proposed than an alternative way of talking about technology use is
required. A stylistic analysis of technologies, drawing on Bakhtin’s analysis of novels, is outlined as one
potential alternative.

Introduction
The concept of ‘affordance’ features frequently in research into technology and learning. The term
is used, in various ways, to explain people’s relationships with technologies. (A fuller review of
these uses will be provided later in the article.) Unsurprisingly, this makes the concept extremely
important for this field – potentially a foundational concept for research (Conole & Dyke, 2004a).
This proposal has been challenged (Boyle & Cook, 2004), but primarily at an operational level –
more critical questions about the concept, such as those asked in the field of human–computer
interaction (e.g. Albrechtsen et al, 2001), remain unasked here.
The purpose of this article is to take a critical position by exploring the history of affordance,
the assumptions underlying it and implications of using it. This will be achieved by tracing the
discourse of affordance from its inception in Gibson’s theory of ecological perception, through the
literature on design, into human–computer interaction and finally to learning and technology.

A History of the Discourse

The Origin of Affordances
Gibson’s book The Ecological Approach to Visual Perception (1979) grew from studies during World
War II into the depth perception of pilots. Frustrated that these failed to predict flight performance,
he devised an alternative theory of perception, proposing an approach that, initially, appears
sympathetic to current situated perspectives on action:
When no constraints are put on the visual system, we look around, walk up to something
interesting and move around it so as to see it from all sides, and go from one vista to another.
That is natural vision, and that is what this book is about. (Gibson, 1979, p. 1)
The theory of ecological perception proceeds by specifying a unit of analysis (‘The words animal
and environment make an inseparable pair. Each term implies the other’, [p. 8]), and concepts are
defined (e.g. mediums, substances and the surfaces that separate them, [p. 16]) that permit an
ecological epistemology to be developed: ‘Because of illumination the animal can see things;

402
The Problem with Affordance

because of sound it can hear things; because of diffusion it can smell things. The medium thus
contains information about things that reflect light, vibrate or are volatile’ (p. 17).
This notion that information is simply present in the ‘ambient optic array’ (the convergence,
at our point of perception, of various diffracted rays of light) is thus used to challenge common
theories of meaning:
The ambient stimulus information available in the sea of energy around us ... is not transmitted,
does not consist of signals, and does not entail a sender and receiver. The environment does not
communicate with the observers who inhabit it. Why should the world speak to us? ... The
world is specified in the structure of the light that reaches us, but it is entirely up to us to perceive
it. (p. 63)
It is this spirit of transparent information that underlies the concept of affordance: ‘The affordances
of the environment are what it offers the animal, what it provides or furnishes, either for good or ill’
(p. 115, original emphases). Gibson, emphasising the ‘complementarity of the animal and
environment’, stresses that affordances are ‘relative to the animal’ (p. 115). This has interesting
consequences:
The affordances of the environment ... are in a sense objective, real and physical. ... An
affordance is neither an objective property nor a subjective property; or it is both if you like. An
affordance cuts across the dichotomy of subjective-objective and helps us to understand its
inadequacy. It is equally a fact of the environment and a fact of behavior. It is both physical and
psychical, yet, neither. An affordance points both ways, to the environment and to the observer.
(p. 129; cf. p. 41)
This claim is partly explained by his epistemological commitments. Specifically and repeatedly, he
attacks the idea that mental activity is involved in perception. (‘These are still the operations of the
mind upon the deliverances of the sense, and there are too many perplexities entailed in this
theory. It will not do, and the approach should be abandoned’ [p. 238; cf. p. xiii, 2].) He dismisses
the Paigetian approach that ‘emphasizes what the children believe instead of what they see’ (p. 195)
and adopts a materialist, essentialist position: ‘Let us suppose that a kind of essential structure
underlies the superficial structure of an array when the point of observation moves. This essential
structure consists of what is invariant despite the change’ (p. 73).
Given this position, how should Gibson’s claim that affordances are neither subjective nor
objective be interpreted? This claim begs the question of what the relationship consists of. There
seem to be few possibilities, and none of them are satisfactory. An initial position might be that the
relationship consists of all the interactions between object and perceiver that are possible. What,
though, is the ontological status of this set of actions? ‘Possible’ leaves nothing substantial to work
with, unless a definitive list of possibilities can be constructed from the properties of each element
separately, and their various interactions. This is only feasible if people and things can be
exhaustively described, which presupposes both an essentialist, positivist epistemology (which is
consistent with Gibson’s position) and a finite number of characteristics (which is not: see p. 243).
Lacking this, ‘possible’ is a work of imagination and ‘affordance’ becomes speculative rather than
analytic. This possibility will be returned to later.
If the nature of this relationship is not speculative, it must be either historical or normative. If
it were historical, all that could be said then is that a thing afforded something to someone in a
specific circumstance, not that it affords it in general. This path of diminishing returns trivialises the
concept analytically. (Chemero [2003] offers this as a solution, but attempts to preserve generality
by arguing that affordances depend upon potential, not actual, observers – inconsistently reverting
to ‘possibilities’.)
Next are normative interpretations (which Chemero sees no problem with), implying things
about ‘typical’ or ‘expected’ interactions – however, ‘typical’ presupposes an historical analysis, and
‘expected’ implies a mental activity of the kind that Gibson wishes to rule out. It also fails to specify
what counts as ‘normal’ (Potter & Wetherell, 1987).
This suggests that, in spite of the claimed relational basis for affordances, they are most
coherently interpreted either as works of imagination or as positivistic properties of objects, albeit
ones activated or perceived by people and animals. They arise from the environment, where the
‘invariants’ that specify them are located. This interpretation is consistent with Gibson’s own

403
Martin Oliver

speculations: ‘Perhaps the composition and layout of surfaces constitute what they afford’ (Gibson,
1979, p. 127).
Ecological psychologists recognise this problem – for example, Chemero (2003, pp. 182-183)
referred to affordances as ‘impossible, ghostly entities, entities that no respectable scientist ... could
have as part of their ontology’.
It is interesting to consider the kinds of claims Gibson makes for affordances. Most are
animalistic or primitive (Gibson, 1971); technology, if mentioned, is stone-age (e.g. clubs, hand
axes, spears, awls [Gibson, 1979, p. 40; cf. p. 128]): ‘All these offerings of nature, these possibilities
or opportunities, these affordances as I will call them, are invariant. They have been strikingly
constant throughout the whole evolution of animal life’ (1979, p. 19).
Gibson does attempt to extend this theory to more complex affordances – but the examples
are sparse and unconvincing (e.g. 1979, p. 42), resorting to what can only be interpreted as visual
stereotypes (e.g. ‘Mates, and amiable animals are distinguished from ... hostile animals by their
shapes, colors, textures, and deformations’ [1979, p. 232]).
He also concedes that ‘information pickup’ (direct, unprocessed perception) is not always
straightforward (e.g. 1979, pp. 152, 157) and that ‘misinformation’ may result in ‘misperception’
(1979, pp. 142, 243). He attempts to explain away this problem by shifting responsibility onto the
inattentive perceiver, rather than his positivistic epistemology or proposed perceptual system
(1979, p. 243). Perceptual problems relating to dreams, drugs or delusions are also the perceiver’s
fault, he argues, since they cannot or will not ‘make the ordinary tests for reality’ (1979, p. 257).
The problems of information pickup thus remain unresolved, practically. This renders affordances
problematic. Again, Gibson can only offer visual stereotypes as a solution (with no mention of how
these are acquired).
The misperceiving of affordances is a serious matter. As I noted in Chapter 8, a wildcat may look
like a cat. (But does he look just like a cat?) A malevolent man may act like a benevolent one. (But
does he exactly?) The line between the pickup of information and the failure to pick up
information is hard to draw. (1979, p. 244)
This analysis is not reassuring for educationalists; nor is the way Gibson talks about information,
knowledge and learning: ‘The term information cannot have its familiar dictionary meaning of
knowledge communicated to a receiver. This is unfortunate, and I would use another term if I could’
(1979, p. 242).
He explicitly discounts the kinds of knowledge educationalists would value, focusing almost
exclusively on ‘knowledge of the environment’, rather than ‘knowledge that comes from parents,
teachers, pictures and books. …This is a different kind of knowledge’ (1979, p. 253). As soon as
affordances are introduced, Gibson appropriates ‘value’ and ‘meaning’ in a similar way (1979,
p. 140) – as environmental properties that can be ‘directly perceived’ (1979, p. 127). This, he alleges,
does away with the need for learning (see, for example, 1979, pp. 128, 40). Gibson tries to explain
other sources of knowledge, such as writing and art:
Pictures and sculptures ... contain information and make it available for anyone who looks. They
nevertheless are, like the spoken and written words of language, man-made. They provide
information that ... is mediated by the perception of the first observer. They do not permit
firsthand experience – only experience at second hand. (1979, p. 63; cf. p. 42)
Elsewhere he attempts to explain how indirect knowledge differs from ‘normal’ ecological
knowledge, differentiating direct perception from mediated knowledge. The latter is still directly
perceived (it still acts as a stimulus), but operates differently; he proposes that ‘perceptual
mediators’ are invented and have to be learnt (Gibson, 1965), but he does not explain how or
provide an account of meaning in relation to mediation. He also does not explain any relationship
between this and affordance.
Indeed, in spite of his attempts to argue the irrelevance of the mind, Gibson repeatedly states
that affordances need learning (e.g. 1979, pp. 143, 194, 235; Gibson, 1971). Even for the most ‘basic’
affordances, he concedes that perception might need to develop in some way – something that
most people would call a kind of learning even if Gibson avoids describing it as such (Gibson, 1979,
pp. 168, 241, 253).

404
The Problem with Affordance

Gibson’s defence involves denying, again, the role of the mental, arguing instead for a process
of ‘attunement’:
The theory of information pickup ... needs to explain learning, that is, the improvement of
perceiving with practice and the education of attention, but not by an appeal to the catch-all of
past experience or to the muddle of memory. The state of a perceptual system is altered when it
is attuned to information of a certain sort. The system has become sensitized. (Gibson, 1979,
p. 254; emphasis added)
Unsurprisingly (given his radical agenda), other theorists of perception have challenged Gibson’s
position. Gregory, for example (1998), presents a position far easier to reconcile with constructivist
perspectives.
Perceptions are predictive hypotheses, based on knowledge stored from the past. ... Perceptions
are not regarded as internal pictures or sounds, but rather as language-like descriptions coded,
we suppose, by brain structures of what may be out there. We carry in our heads predictive
hypotheses of the external world of objects and of ourselves. (Gregory, 1998, p. 1693)
Gregory’s position is not actually as far from Gibson’s as might at first be supposed:
The perceptual process is not primarily one of supplementing or contracting but of selecting. If so,
the contribution of the perceiver is only one of paying attention, of looking, of exploring, of
adjusting the eyes, and of detecting the ‘formless invariants’. (Gibson, 1965)
The crucial difference is that Gibson fails to account for how this selecting takes place; the
intentionality of selection, recognised by Gregory as knowledge-informed, remains mysterious in
Gibson’s scheme.
Other commentators have attempted to redeem Gibson’s position through clarification.
The existence of the affordance is independent of the actor’s experiences and culture, whereas
the ability to perceive the affordance may be dependent on these. Thus, an actor may need to
learn to discriminate the information in order to perceive directly. In this way learning can be
seen as a process of discriminating patterns in the world, rather than one of supplementing
sensory information with past experience. (McGrenere & Ho, 2000, p. 2)
But this does not resolve the situation. This distinction contradicts Gibson’s assertion that
perception is ‘direct’ and yet still runs into the problem of what the ‘relationship’ between actor
and environment consists of. It also renders affordances unknowable – a problem that will be
returned to in the next section.
To summarise, Gibson advanced a radical agenda for perceptual psychology – one that is
situated, but anti-‘mentalist’. Affordances are positioned as relational, but Gibson failed to explain
what this relationship consists of. Although he tries to argue that perception is direct, Gibson
cannot avoid talking about learning (sometimes under pseudonyms). Whether or not affordances
exist, perceiving them is problematic – they are relevant ontologically, but not epistemologically.
Although the idea of affordance might be explicable as a normative account, and thus acceptable at
the ecological level of species (e.g. justified through ‘natural selection’; Chemero, 2003), this is of
little comfort to educators, concerned with specific people or groups learning things.

Norman and the Adoption of Affordances for Design

Like Gibson, Norman embarked on a radical and passionate project – The Psychology of Everyday
Things (POET; Norman, 1988) – because of personal frustration; in this case, with poor designs
(Norman, 1988, p. vii). Central to this book is the appealing idea that designers should focus upon
users’ perceptions, conventions and errors rather than valuing aesthetics over usability (1988,
pp. 150-151).
He advocates two things (p. 12) as ‘fundamental principles of designing for people: (1) provide
a good conceptual model and (2) make things visible’. Given this centrality of visibility to his thesis
(p. 4), it is unsurprising that Norman adopted ideas he had discussed with Gibson when both
scholars were at La Jolla (Norman, 1999).

405
Martin Oliver

Affordances provide strong clues to the operations of things. Plates are for pushing. Knobs are for
turning. ... When affordances are taken advantage of, the user knows what to do just by looking:
no picture, label or instruction is required. (Norman, 1999, p. 9)
However, Norman disagreed with Gibson about whether the mind processed or simply ‘picked up’
information (Norman, 1999; see also Norman, 1988, p. 219; 1969). Consequently, he introduced a
subtle but important change into the way he used the term: ‘The term affordance refers to the
perceived and actual properties of the thing, primarily those fundamental properties that determine
just how the thing could possibly be used’ (Norman, 1988, p. 9).
He later expanded on the difference he intended:
POET was about ‘perceived affordance’. When I get around to revising POET, I will make a
global change, replacing all instances of the word ‘affordance’ with the phrase ‘perceived
affordance’. The designer cares more about what actions the user perceives to be possible than
what is true. (Norman, 1999, p. 21)
This shift solves the problem of specifying what a thing affords. However, his user-centred
alternative creates other problems. Firstly, ‘affordance’ becomes redundant as an analytic concept.
Norman clarified: ‘There can be both real and perceived affordances, and the two sets need not be
the same’ (Norman, 1999, p. 24).
The implication of this, however, is that ‘real’ affordances are unknowable. If any given set of
perceived affordances may be different to the ‘real’ one, how can we ever know which set is ‘right’?
All we have access to is what we can perceive – thus all we can ever access are ‘perceived’
affordances. Norman does not seem to recognise this problem, let alone resolve it; he sees no
inconsistency in his interpretative position and the positivist claim: ‘Affordances reflect the possible
relationships among actors and objects: they are properties of the world’ (Norman, 1999, p. 10).
There are related problems. To explain these, it is necessary to introduce other concepts from
POET. Norman differentiates between affordances, constraints and mappings (Norman, 1988,
p. 12). Constraints may be physical, semantic, cultural or logical. He proposes that physical
constraints can be used to prevent certain operations and that ‘desired actions can be made
obvious, usually by being especially salient’ (p. 84). Oddly, since ‘obviousness’ implies
interpretation, he does not position this as a semantic constraint. Indeed, he has remarkably little to
say about these, other than: ‘Semantic constraints rely upon our knowledge of the situation and of
the world. Such knowledge can be a powerful and important clue’ (p. 85).
Arguably, these could be incorporated into the next concept: cultural constraints. These he
describes as ‘conventions’ and reduces to cognitive schemas or frames (pp. 84-85). He recognises
them as significant sources of error but has little to say about how to design with or to overcome
them. In spite of recognising their utility, he rails against their arbitrariness, since there are all too
often situations in which such conventions fail to hold (e.g. p. 168).
Interestingly, as an example, he states, ‘one cultural convention is that signs are meant to be
read’. This is important, given his commitment to perceived affordances. Gibson’s affordances are
supposed to cross the subjective/objective binary, but this contingency is a concession that
interpretation (even at the level of engagement with a design) is learnt, not something ‘in’ the
artefact. This contingency arising from interpretation is compounded where he discusses how
people ‘explain away’ errors or ignore their beliefs due to peer pressure: ‘Everyone forms theories
(mental models) to explain what they have observed. ... People are free to let their imaginations
run free as long as the mental models they develop account for the facts as they perceive them’
(p. 39). They have no way of telling which account is ‘true’ – even hindsight is merely justifying
through post hoc rationalisation (pp. 128-129).
What is left, then, of the supposed objectivity of affordances? All we can work with is
socialisation and learning. Even Norman’s discussion of logical constraints rapidly returns to
‘natural mappings’ that are actually conventional. Interestingly, that ‘naturalisation’ is central to
Norman’s thesis is illuminating. To position something as natural is to render it invisible,
unquestionable; to imply that it could not be otherwise (Potter & Wetherell, 1987). Consider, then:
‘The vertical plate and supporting pillars are natural signals, naturally interpreted, without any need
to be conscious of them. I call the use of natural signals natural design and elaborate on the
approach throughout this book’ (Norman, 1988, p. 4). He goes on to claim that car design is

406
The Problem with Affordance

‘natural’ (p. 22), revealing that he means something specific by this: ‘Natural mapping, by which I
mean taking advantage of physical analogies and cultural standards, leads to immediate
understanding’ (p. 23).
Analogies (even physical ones) are clearly interpretative, the product of socialisation, leaving
‘nature’ as a symptom of culture. He further differentiates between ‘natural’ (physical) properties
and ‘artificial’ (cultural) conventions, criticising the latter for being difficult to learn, since rote
learning (‘the bane of modern existence’, p. 68) gives no hint about the underlying conceptual
model. He does, however, recognise such ‘arbitrary’ conventions as a useful last resort: ‘When
something can’t be designed without arbitrary mappings and difficulties, there is one last route:
standardize. ... The nice thing about standardization is that no matter how arbitrary the
standardized mechanism, it has to be learned only once’ (1988, p. 200). This all defines ‘good’
design culturally, not in terms of properties of the world.
Furthermore, he explains how a ‘good’ mental model ‘transforms an apparently arbitrary task
into a natural one’ (1988, p. 69). This suggests that personal meaning is key, for Norman, and that it
can even be created for arbitrary designs. The quality of design thus lies not in its substance, but in
the ability of the user to read it – or to imagine how to read it. This is made clearer in a section
where he discusses the role of three conceptual models in design:
Three different aspects of mental models must be distinguished: the design model, the user’s model
and the system image. ... The design model is the conceptualization that the designer has in mind.
The user’s model is what the user develops to explain the operation of the system. Ideally, the
user’s model and the design model are equivalent. However, the user and designer communicate
only through the system itself: its physical appearance, its operation, the way it responds, and the
manuals and instructions that accompany it. Thus the system image is critical: the designer must
ensure that everything about the product is consistent with and exemplifies the operation of the
proper conceptual model. (1988, pp. 189-190)
This passage is useful in re-casting design as communication. It highlights the importance of
making intentions – meaning – clear; it re-positions the user as a reader and the designer as an
author, with design as a form of semiotics. The problems of design become social ones of sharing
meanings. This sense of design is echoed throughout the book (e.g. 1988, pp. 2, 198). Thus to take
advantage of the ‘affordances’ of things, users must learn to interpret their design, based either on
past experience or on information presented to us (1988, p. 82).
Norman’s assertion that ‘affordances…convey messages about their possible uses, actions,
and functions’ is clearly at odds with Gibson’s position that the world does not communicate with
us (Norman, 1988, p. 242). Indeed, so little of Gibson’s intended sense of the word remains that the
appropriateness of its use must be questioned. Moreover, as McGrenere & Ho point out (2000,
p. 3), Norman has conflated affordances with the information specifying affordance; as argued
earlier, given the problematic (and unknowable) nature of affordances per se, either falling back on
species-level generalisations or a retreat to learnt responses seem to be the only viable courses of
action.
Norman thus leaves himself in an untenable position. If affordances are subjective and
interpretative, they cannot form an unambiguous basis for design. He is left in the limited position
of only considering ‘normal’ users – those who have already learnt simple responses to things.
(Gibson’s position is similar, in that species-level similarities are frequently assumed.) He also
assumes, patronisingly at times, that complex things may be too hard to learn (e.g. an application
sounding remarkably like Microsoft Word being criticised because ‘there is no way that a program
can remain usable and understandable by the time it has all of those special-purpose features’ [1988,
pp. 172-173; cf. p. 213]). Norman does recognise that people’s learning is heterogeneous,
differentiating (for example) between designers and ‘typical users’ and discussing deterioration of
ability with age (1988, pp. 151, 155, 162-163).
Thus, focusing on the role of learning and cultural contexts offers a new way to understand
the point of POET. If ‘natural’ is understood as ‘taken-for-granted culturally’, we end with a
position that is consistent and educationally relevant – but nothing to do with Gibson’s affordances.
Norman may have introduced affordances to design, but what remains valuable in his position is
the role of cultural conventions, which he was so critical of.

407
Martin Oliver

Affordances in Human–Computer Interaction

Given Norman’s explicit discussion of computers and design, it is hardly surprising that his version
of ‘affordance’ was taken up by human–computer interaction (HCI) researchers. McGrenere & Ho
(2000) have provided a review of uses of the term, identifying several parallel conventions:
• Gibson’s notion of an action possibility or offering;
• Norman’s notion of a perceived suggestion;
• As an interface object; and
• In unclear or inconsistent ways.
They suggest ‘affordance’ is often understood as ‘the design aspect of an object which suggests how
the object should be used’ (2000, p. 1) because Norman collapsed the utility of an object with the
perception of that object’s utility (2000, p. 3).
McGrenere & Ho advocate a return to a developed version of Gibson’s formulation. Two
developments are proposed (2000, p. 2): analysis of complex affordances that arise from hierarchies
of simpler affordances, and also analysis of degrees of affordance (focusing on minimising effort),
rather than a binary scheme. They also distinguish the utility of an object (Gibson’s sense) from its
usability (Norman’s sense) in a way that is productive for designers. Similarly, Torenvliet (2003,
p. 14) discusses how affordance in Gibson’s sense can be useful ‘because it forces us to think more
than screen deep’, considering how ‘clicking’ an icon actually entails a physical mouse button press
leading to a series of state changes within the computer. (He also criticises Norman’s influence,
leading HCI researchers to only consider ‘perceived’ affordances, effectively reversing the
definition of the term so that it indicates two contrary meanings.)
Albrechtsen et al (2001), who also reviewed this field, argue that both Norman and Gibson
are ‘considering affordances as more or less static surface phenomena’ (p. 10). They point out that
neither really address high-level cognitive issues such as culture, language or knowledge (p. 15);
they thus consider ‘higher order affordances’, linked to Wittgenstein’s idea of language games.
They do highlight points of continuity with Gibson, such as the focus on relationships, rather than
tool or person, although their situated analysis arguably sidesteps the problems of generalisation
faced by Gibson. They take an interest in tools as ‘extensions of the body’ (in Gibson’s terms),
drawing an analogy with the functional organs described by Leont’ev. Perhaps this consonance is
partly because they share a positivistic foundation with Gibson, discussing (for example) the
properties of objects or the way that displays present (rather than construct) the ‘states of the
system and the environment’ (p. 26).
Their position is, however, quite a ‘development’ – they study mediated understanding
(contrasted with unmediated direct perception), deny the timelessness of affordances (posing
instead an historical analysis), are less concerned with ‘natural’ affordances (focusing on designed
artefacts) and argue that affordances are contingent on the situation, actors and purposes (not
essential environmental properties). They also, somewhat dubiously, talk about direct perception
of ‘the deep structure of the work’, asserting (presumably metaphorically) that social structures
provide invariants that specify affordances of the work domain. This radically redefines affordances
as ‘cues for action relevance’ (p. 27; making them firmly interpretative) or even as ‘social
intentions’ (p. 30). Again, although their position is interesting and addresses many of the problems
identified above, this is achieved by turning Gibsons’s concept of affordances into re-branded
Activity Theory. Quite why the re-branding is necessary, when Activity Theory already seems
adequate, is unclear.

Affordances and Learning with Technology

The overlap between HCI and e-learning has led to affordances being taken up by researchers in
this area, too. Like HCI, use of the term ‘affordance’ is variable. For example, Laurillard et al (2000)
assert: ‘Examples for education would be: A large lecture affords listening. A small group also
affords preparing to speak. In each case the features as perceived by an observer create the
possibility for a certain kind of behaviour’ (p. 3).
The first example suggests Gibson’s concept. The second does not involve action as
understood from an ecological perspective. (Preparing to speak, rather than speaking, is arguably a

408
The Problem with Affordance

private, ‘internal’ process, failing to meet the requirement of affordances as related to animal–
environment pairings.) The explanation, focusing on what is perceived, suggests Norman’s use of
the term. Although it might be possible to interpret all of these as perceived affordances, making no
claims about what is ‘offered’, it is Gibson, not Norman, who is referenced.
Particularly useful in illustrating this tradition are the discursive articles provoked by Conole
& Dyke’s proposal to use affordances as the basis for research (2004a). Their definition is derived
from Norman by way of Salomon. However, although they assert (2004b, p. 301) that by this they
mean to focus on ‘“possible” use’, they discuss a negative sense in which affordances of objects
‘often restrict their accessibility to users’ (2004a, p. 115). Equally, although they clarify that they are
interested in the relationship between technological infrastructure and people’s use of those
technologies, they are still concerned with the ‘properties of a thing’ (albeit those that are perceived
as well as those it possesses), once again committing to a positivist, essentialist position. (Even in
the subsequent clarification, qualities such as ‘speed of change’ are described as ‘an inherent and
important characteristic of technologies’ – 2004b, p. 304.) This position is critiqued by Boyle &
Cook (2004), who point out the discrepancy between the espoused social basis of Conole & Dyke’s
analysis and Gibson’s explicitly unsocial, non-constructivist position.
Conole & Dyke then draw on ‘relevant social theory and critique’ to create ‘a list of the
potential themes and commonalities ... distilled into a taxonomy of ICT [information and
communication technology] affordances’ (2004a, p. 116). This list groups together qualities
attributable to the technology (e.g. multimodality and non-linearity), to its user (e.g. reflection) and
to their mutual relationship (e.g. immediacy). Arguably, such a list does not conform neatly to
Gibson’s, Norman’s or McGrenere & Ho’s formulation of affordance. Some elements seem
consistent with the essentialised, positivist origins of affordance. Others seem entirely unrelated –
reflection, for example, would be denounced by Gibson. Moreover, the idea that reflection might
be a response to an offering by technology (implying some causal link) rather than an act of
personal agency seems odd. Additionally, Conole & Dyke assert that they ‘are also interested in
exploring the creative and innovative way that people respond to technologies and perhaps adapt
them for use in unforeseen circumstances’ (2004b, p. 301), rather than just considering designers’
intentions. This recognises earlier critiques of the predictive value of Gibson’s affordances but does
not resolve them; there is still no way of identifying these possibilities or specifying what they
consist of.
Indeed, there seems to be no unifying concept behind the list; elements may emerge from the
literature, but the notion of ‘affordance’ seems ill-suited to legitimating this conglomeration of
claims about perceptions, actions and characteristics. Something much broader is required.
Substituting the word ‘claims’ for ‘affordances’, for example, provides a more plausible framework
with no loss of the central message and no diminution of utility to practitioners. For example:
Many of the [claims] outlined above are known intuitively, but the development of an explicit
taxonomy provides a basis for discussion, critique, and further refinement. This may involve
questions such as:
– Are these the only [claims]? …
– Can particular [claims] be mapped to particular types of technologies? (Conole & Dyke, 2004a,
p. 121)
Repositioning the taxonomy as a socially constructed account bypasses the ontological confusion
and inconsistency inherent in the current formulation – although it does not confer the status that
Gibson’s oft-cited, 25-year old text might. It also helps explain Boyle & Cook’s criticism (2004,
p. 297) that the paper contains ‘a certain amount of hopeful expectation that affordances and
abilities will simply emerge’, since these themes are emergent, having been synthesised from
research narratives.
Boyle & Cook suggest alternatives. One is to extend the ‘ecological’ approach underlying
Gibson’s thesis so as to focus upon how technology has changed the ‘habitat’ in which we live. This
suggestion echoes the ‘information ecologies’ proposed by Nardi & O’Day (1999), but ignores the
issues facing Gibson’s formulation of ‘affordance’ and also its primitive and ‘strikingly constant’
(Gibson, 1979, p. 19) scope. Gibson does not consider the idea that an animal might develop its
‘niche’ (the collection of affordances that make a particular ecological setting a suitable habitat).
Nonetheless, if the other issues in Gibson’s thesis (especially the attitude towards the mind) were

409
Martin Oliver

also addressed, this proposal may be worth pursuing – although it remains interpretative, rather
than helping to identify the possibilities for action ‘afforded’ by technology that prompted interest
in the term.
The second suggested alternative focuses on ‘social affordances’, which are interpreted as
properties incorporated into computer-supported collaborative learning environments that ‘act as
social facilitators’ (Boyle & Cook, 2004, p. 298). ‘When they are perceptible, they invite the learner
to act in accordance with the perceived affordance, i.e., start a task or a non-task related interaction
of communication’ (Kreijns et al, cited in Boyle & Cook, 2004, p. 298). Examples include tools for
monitoring communal activity. Such tools might be useful, but they miss the point of affordances –
they extend what is noted (by supporting the perceptual process with more visible signals) but they
do not ‘offer’ new possibilities for action; they are simply intended to encourage particular extant
social activities. Here, the extension of Norman’s use of the term becomes particularly misleading,
confusing usability and utility.

Discussion
Use of the term ‘affordance’ diverged after Norman appropriated it. One line of development is
consistent with Gibson’s original position, although arguably these refinements have not countered
problems with its relevance to learning. This tradition is positivistic and essentialist, at odds with
contemporary educational thought. This is, perhaps, inevitable since Gibson seems preoccupied
with ontology, not learning, and explicitly sought to rule out the ‘“subjective” world,’ or the world
of ‘“consciousness”’ (Gibson, 1979, p. 129). Arguably, it would be inappropriate to advocate a
return to this position in educational research.
Jones (2005) argues that it is possible to overcome these difficulties by emphasising the non-
dualist, realist and relational nature of affordances. However, this position does not address either
of the two main critiques offered. No ontological explanation is provided of what this relationship
between animal and environment consists of. (Is it historical, imagined or something else? Even if it
is one of these, the problem of whether this resolution has implications for predictive analysis still
remains – this prescription of possible interactions being the hope that gives the concept its analytic
appeal.) Second, the assumption that a totalising ‘view from nowhere’ can be inferred from specific
perceived instances fails to account for the idea that affordances might exist but never be perceived,
and is also an epistemologically suspect inferential move (cf. Albrechtsen et al, 2001).
The second major tradition builds on Norman, using the term in a looser sense. The analysis
above suggests that Norman’s position may have consistency, but not for the reasons he proposed.
If design is reinterpreted as a form of cultural literacy this position addresses the predictive
requirements that are sought, but affordances have no analytic role to play in this position, and
indeed in their original sense are inconsistent with it (cf. Albrechtsen et al, 2001). To adopt this
position, arguably, the term ‘affordance’ should be abandoned as being misleading since its original
connotations cannot be left behind.
A third position is suggested by Boyle & Cook (2004), which might involve developing the
ecological approach further. It may be that Nardi & O’Day’s (1999) work on information ecologies
contributes to this, but since affordance is not a major part of that work, this approach will not be
pursued further here.
The problem with these divergent lines of development is that ‘just as printing new money
devalues existing money, the more new definitions the term affordance gains, the less value any
one of them has’ (Torenvliet, 2003). Where, then, does this leave researchers and designers?

I have No Affordances, Yet I Must Design

Arguably, researchers and designers persist in using the term affordance because of the desire to
speak about technology in a general enough way that it can be theorised in its own right. It also
promises an appealing control over users – if affordances permit some actions and constrain others,
users are more likely to behave as expected. There is thus a tension between recognising the
complexity of the concept and having a language simple enough to be useful.

410
The Problem with Affordance

Does it matter, however, if the term is used inconsistently, if it is productive? Arguably, yes.
For researchers, incoherence in terminology will provide a false impression of research coherence,
hiding (not resolving) problems. For designers, the concept might prompt reflection, but if that
reflection underplays the role of the user (Norman’s critique of Gibson) or of learnt cultural
conventions (the critique of Norman advanced here), then increasing its quantity, rather than its
quality, is a disservice to users. Some coherent, plausible alternative is needed.
There are at least two methodologically and epistemologically coherent ways in which this
tension can be met. The first extends the claims-based approach used (not espoused) by Conole &
Dyke. Working with accounts of technology (reported experience, research papers and so on.)
brings the discussion, epistemologically, to a single level: discourse. These discourses can be
identified, genealogies constructed, their use and conflicts studied. Methodologies for this exist,
typically drawing on Foucault’s work.
However, some may find this retreat from the object (technology) to talk about the object
unsatisfying. There may be an alternative, although its viability is yet to be demonstrated.
The analysis above of designed artefact as cultural text suggests that it may be possible to
undertake a literary analysis of technology. Studying the stylistics of technology may permit it to be
talked about without resorting to a positivisitic position. (This perspective is not entirely
unprecendented; Albrechtsen et al, 2001, talk about artefacts as ‘texts’ or ‘narratives’ arising from
research into intuitive interfaces [p. 12].)
Bakhtin’s difficulty in studying the novel is analogous in many ways to the situation here:
The study of the novel as a genre is distinguishable by particular difficulties. This is due to the
unique nature of the object itself: the novel is the sole genre that continues to develop. ... The
forces that define it as a genre are at work before our very eyes: the birth and development of the
novel as a genre takes place in the full light of the historical day (Bakhtin, 1987a, p. 3).
There is particular merit, as well as difficulty, in studying such evolving forms. Bakhtin argues (p. 7)
that as it develops ‘it reflects more deeply, more essentially, more sensitively and rapidly, reality
itself in the process of unfolding’. It thus provides an important lens through which to study culture
more broadly; in this case, studying technology illustrates issues about education more widely.
This does not require a simple and positivistic position, but sees ‘reality’ as social and
historical (Bakhtin, 1987b, p. 263). Part of this position – one that offers interesting but relatively
neglected possibilities for the study of technology and education – is the assertion that all language
use is ideological (Bakhtin, 1987b, pp. 270-271). Additionally, the novel is presented as reflecting
heteroglossic language use. Studying the way in which technologies are brought together to
automate practices (such as surveillance, communication and so on) – practices that may or may
not cohere successfully – promises to reveal the ideological positioning of particular systems,
designs or devices through a reading of the elements that they consist of. This would allow
technology to be described in a way that recognises its social and historical production, as well as
intentionality, without proposing simple causal relationships.
Importantly, Bakhtin positions the ideology of language as dialectic, not transmissive.
Speakers can appropriate ‘alien’ expressions, making them their own (Bakhtin, 1987b, p. 282). This
echoes Conole & Dyke’s (2004b) desire to recognise creative uses of technology, beyond the
intended use. However, it does so analytically, exploring the intentionality of various actors
(Bakhtin, 1987b, p. 292). There is no recourse to a speculative set of possibilities, but instead the
analysis of concrete instances of expression – something as coherent as the previously proffered
analysis of discourses, but maintaining a focus on technology. This resolves the difficulty of
specifying the nature of the relationship between technology and people. It does not claim to solve
the problems of prediction facing affordances, although it is coherent with an approach that
involves studying existing repertoires of use and recognising the potential of analogous or newly
imagined applications.
Methods for undertaking such work have yet to be developed, but this analysis suggests that
such work may be productive.

411
Martin Oliver

Conclusions
This article calls into question whether the concept of ‘affordance’ has any analytic merit for
researchers or designers. The critiques advanced here suggest that its value is limited unless we are
willing to abandon constructivist values in order to explore ‘inherent properties’ in a positivistic
sense. The concept either allows us to talk about technology as part of the environment (Gibson’s
sense), but not about minds; or is rendered analytically ineffectual by recognition of the cultural,
constructive nature of learning (Norman). The only developments that seem to escape this
dilemma (e.g. Albrechtsen et al) do so by keeping the label but effectively abandoning the concept
(e.g. by it becoming Activity Theory). On the one hand, we abandon current conceptions of
learning; on the other, we effectively abandon affordance – although researchers seem remarkably
reluctant to give up the term.
Even in Gibson’s sense the term is problematic. It may be adequate for discussions of species
and generalities, but people and learning are beyond it. Furthermore, the need for people to learn
in order to take advantage of affordances is unexplained and Gibson fails to provide any analytic
way of identifying affordances. His discussions resort to commonsense examples that are not
exhaustive or definitive, merely illustrative. This does not give a method for research or design, and
led to Norman’s veiled reversion to designing around cultural norms. By conceding that
affordances need to be learnt but failing to explain their relationship to ‘attunement’, Gibson leaves
himself vulnerable to the same charge that he levelled at cognitive processing theorists (1979,
p. 238): there are too many perplexities, and thus this explanation should be abandoned.
His position gives primacy to environment over people and only recognises people as tool
using, not tool making; it ignores our agency. It may be more productive for this field to focus not
upon the ‘offered’ possibilities but upon what a person imagines might be possible – and also upon
what they can imagine doing to achieve the same end with some other object. If an object does not
allow us to undertake an action, we can find an alternative or make a new tool that does.
Norman’s re-interpretation seeks to recognise some of these issues, but by doing so he
renders the notion of ‘real’ (as opposed to ‘perceived’) affordances irrelevant. This has led to the
current position in which the term is used in a degenerate way. Since these different traditions
cannot be eradicated, this confusion will persist, muddling both research and design; arguably,
then, the term ought to be abandoned. Nevertheless, the coherent elements of his position can be
reinterpreted in a way that has promise. Using literary techniques to study how people ‘write’
(design, create) and ‘read’ (use and apply) technology may support research that is coherent and
socially aware without trivialising the complexity of this field of study by resorting to an
essentialist, positivist position. We should just avoid calling what we study ‘affordances’.

References
Albrechtsen, H. Andersen, H., Bødker, S. & Pejtersen, A. (2001) Affordances in Activity Theory and
Cognitive Systems Engineering. Internal report, Riso National Laboratory, Denmark.
Bakhtin, M. (1987a) Epic and Novel: towards a methodology for the study of the novel, in M. Holquist (Ed),
The Dialogic Imagination: four essays, pp. 3-40. Austin: University of Texas Press.
Bakhtin, M. (1987b) Discourse in the Novel, in M. Holquist (Ed), The Dialogic Imagination: four essays,
pp. 259-342. Austin: University of Texas Press.
Boyle, T. & Cook, J. (2004) Understanding and Using Technological Affordances: a commentary on Conole
& Dyke, ALT-J, 12(3), pp. 295-299.
Chemero, A. (2003) An Outline of a Theory of Affordances, Ecological Psychology, 15(2), pp. 181-195.
Conole, G. & Dyke, M. (2004a) What are the Affordances of Information and Communication Technologies?
ALT-J, 12(2), pp. 113-124.
Conole, G. & Dyke, M. (2004b) Understanding and Using Technological Affordances: a response to Boyle &
Cook, ALT-J, 12(3), pp. 301-308.
Gibson, J. (1965) The Comparison of Mediated Perception with Direct Perception, Purple Peril of J. J. Gibson,
unpublished manuscript.
Gibson, J. (1971) A Preliminary Description and Classification of Affordances, Purple Peril of J. J. Gibson,
unpublished manuscript.

412
The Problem with Affordance

Gibson, J. (1979) The Ecological Approach to Visual Perception. Boston: Houghton Mifflin.
Gregory, R. (1998) Brainy Mind, British Medical Journal, 317, pp. 1693-1695.
Jones, C. (2005) Who are You? The Experience of Working through an Avatar, E-Learning, 2(4), pp. 414-425.
Laurillard, D., Stratfold, M., Luckin, R., Plowman, L. & Taylor, J. (2000) Affordances for Learning in a Non-
Linear Narrative Medium, Journal of Interactive Media in Education, 2. www-jime.open.ac.uk/00/2
McGrenere, J. & Ho, W. (2000) Affordances: clarifying and evolving a concept, proceedings of Graphics Interface
2000, Montreal, May.
http://www.graphicsinterface.org/cgi-bin/DownloadPaper?name=2000/177/PDFpaper177.pdf
Nardi, B. & O’Day, V. (1999) Information Ecologies: using technology with heart. Boston: MIT Press.
Norman, D. (1969) Memory and Attention: an introduction to human information processing. London: Wiley &
Sons.
Norman, D. (1988) The Psychology of Everyday Things. New York: Basic Books.
Norman, D. (1999) Affordances, Conventions and Design, Interactions, May, pp. 38-43. www.jnd.org
Norman, D. (2002) Affordances and Design. http://www.jnd.org/dn.mss/affordances-and-design.html
Potter, J. & Wetherell, M. (1987) Discourse and Social Psychology: beyond attitudes and behaviour. London: Sage.
Torenvliet, G. (2003) We can’t Afford it! Interactions, July/August, pp. 12-17.

MARTIN OLIVER is a senior lecturer in the London Knowledge Lab, a research centre exploring
the relationship between learning and digital technologies. He is the course leader for the MA in
ICT in Education. Having started off researching how to teach with technology, his research
interests have gradually regressed through supporting people teaching with technology, working
out how to understand what people are doing with technology (methodologies) and now, what
technology is supposed to be anyway. Correspondence: Martin Oliver, London Knowledge Lab,
Institute of Education, University of London, 23-29 Emerald Street, London WC1N 3QS, United
Kingdom ([email protected]).

413