This summer, I had the pleasure of studying under Andrew Wilson at Leeds Beckett University. I really respect Andrew's work (check out his blog https://psychsciencenotes.blogspot.com/), and alongside his PhD supervisor Geoffrey Bingham, their research and approach have really shaped my thoughts on how we should tackle the problem of understanding human behaviour. Two weeks into my time at LBU, Andrew got me to write an essay on the concept of affordances in Ecological Psychology. Keep in mind that this essay wasn't written with the blog in mind (I only started the blog a week later) and was instead written with the purpose of organising my thoughts on affordances. Naturally, there may be parts that seem to assume knowledge on the part of the reader, but I thought it'd be better to get the ball rolling and post the original draft rather than rewriting it. So here we are!
Introduction
The concept of affordances plays a central role in Ecological Psychology theory. But what exactly are affordances? Why do we need them in the first place? How do they help us perceive and act? And how do scientists go about studying affordances? These are some of the questions that I will be attempting to answer, as I wrap my head around and try to articulate the complex nature of affordances.
A first pass
Colloquially speaking, affordances are the possibilities or opportunities for action that are granted by the environment to an organism. For example, a flight of steps affords the affordance of climb-ability; it provides an organism with the possibility of climbing. For the same environment set-up, different affordances will also be made available depending on the abilities of the organism. The same flight of stairs might afford climb-ability via upright walking for an adult, but afford climb-ability via crawling for a baby. At the moment, I’d argue that this is a good enough description of affordances for those who might be getting into it for the first time – it defines affordances in terms of action, and highlights the complementary nature between environment and organism in actualising these actions. However, there is still much to be desired from this definition if we’d like to get to the essence of affordances and the exact role it plays in human behaviour. But before this, we have to understand the reason behind its emergence in psychology.
Why affordances?
The current state of psychology can be historically traced all the way back to the first attempts at studying visual perception, where scientists tried to explain how we came to see the world as it is. The theory that survived was that of the retina image, where light rays emanating from a source would be reflected upon an object before entering our eyes, thereby casting an image of the object onto the back of our retinas, just like how an image would be projected upon an opaque projector screen. There was a huge problem, though – the physics of optics and physiology of our lenses meant that this image was flipped vertically and horizontally, and also two-dimensional. On the contrary, our experience of the world was that of upright and three-dimensional objects. How do we go from images to objects?
To solve this, early scientists suggested that there was something in our minds that recognised this grossly ambiguous image, and through a series of computational steps, fixed the retina image to give it its experienced meaning. This ‘something’ was referred to as the central executive, and it was not without problems. For instance, how did the central executive know that the image was wrong and needed to be fixed? Why didn’t it just assume that this was how the world really was? And even after this, how did the central executive know how to fix the meaningless image? Scientists like Helmholtz argued that the central executive utilised past experiences of the world as a template to fix these vague inputs, but then we would ask, where did these past experiences come from? Given that current traditional approaches to psychology all stem from this Helmholtzian program, they incur what is known as a loan of intelligence that can’t be repaid – we simply can’t explain where and how this initial knowledge (to guide the fixing of subsequent sensations) came to be.
In general terms, the above approach assumes that what is sensed is inherently meaningless, and to which meaning is added through a process of enrichment by the central executive. The many issues of this account led Gibson to assume that meaning is already inherent in the object of perception (i.e., the environment!). This means that perception changes from a two-step process (of sensation and then enrichment) to a one-step process where organisms are simply required to detect the meaning-laden properties of the environment. Given another of Gibson’s assumptions that the primary goal of perception is for action, we take these meaning-laden properties to be related to possibilities for action, or otherwise, affordances!
What are affordances: dispositions or relations?
So, we’ve established the need for assuming affordances exist. But in what way do they exist, exactly? There are at least two accounts on the ontology of affordances. Turvey (1992) was the first to formalise affordances as dispositional properties of the environment that allowed for action to be performed. The complementary action-relevant dispositional property of the organism was termed effectivities (i.e., the abilities and size of the organism). Together, the affordance and complementary effectivity combined to result in the actualisation of the behaviour. For example, according to this account, the pavement has the affordance of being walk-on-able (in that it has an intrinsic property that grants it the possibility of being walked on), while an adult human might have the effectivity of walking (it has the dispositional property that grants it the ability to walk). Put together, we get a system with a relational property of adult-human-walking-on-pavement, that is, the manifestation of the actual walking behaviour! To the extent that this is a relational property is debatable – typically within philosophy, events and occurrences aren’t construed to be properties, so this ‘property’ might very well just be an occurrence that emerges from the adult-pavement system.
On the other side, there are those who instead view affordances as existing by virtue of the relation between animal and environment. For example, Stoffregen (2003) argues that affordances are relational properties that emerge from the animal-environment system, whereas Chemero (2003, 2009) posits that affordances aren’t even properties, but just relations within this system. At least in Stoffregen’s case, the idea is that the environment possesses some physical property, while the animal possesses some ability-related property. Combining these two properties then gives rise to a higher-order property that is the affordance. Using the example from above, the affordance of walk-on-ability is a result of the interaction between (say) the rigidity of the pavement and the adult human’s ability to walk. The given behaviour then necessarily occurs when the affordance co-occurs with the appropriate intentionality.
For me, the above analysis doesn’t make much sense. I’m still thinking and learning about this, so I can’t pretend that I can articulate it formally, but a reading of Mumford (1998) on dispositions defends what he calls a disposition-categorical distinction. Roughly, he argues that both dispositional and categorical properties are both categorical with a dispositional nature. The only way they differ is the kind of dispositional nature involved. More relevantly here, different categorical properties can interact to form a higher-order dispositional property. For instance, sugar has the dispositional property of solubility, and the categorical property of having a certain type of lattice structure. This categorical property (combined with other categorical properties) of sugar is what gives it the dispositional property of solubility. Reflected upon Stoffregen’s analysis, there is simply no need to confer the affordance property onto the animal-environment system. The environment properties he proposes might simply be analogous to Mumford’s categorical properties, a combination of which gives rise to a dispositional affordance property that still belongs to the environment, not the system.
I’m not as familiar with Chemero’s (or even Rietveld’s) affordances-as-relations analyses yet, but I suspect they might suffer from similar issues. It seems like the affordance-as-relations argument largely came about from a) the vagueness of Gibson’s initial definition of affordances and b) a general discontent with the affordance-as-dispositions analysis. Briefly, I point out one alleged issue with the dispositional account:
Colloquially speaking, affordances are the possibilities or opportunities for action that are granted by the environment to an organism. For example, a flight of steps affords the affordance of climb-ability; it provides an organism with the possibility of climbing. For the same environment set-up, different affordances will also be made available depending on the abilities of the organism. The same flight of stairs might afford climb-ability via upright walking for an adult, but afford climb-ability via crawling for a baby. At the moment, I’d argue that this is a good enough description of affordances for those who might be getting into it for the first time – it defines affordances in terms of action, and highlights the complementary nature between environment and organism in actualising these actions. However, there is still much to be desired from this definition if we’d like to get to the essence of affordances and the exact role it plays in human behaviour. But before this, we have to understand the reason behind its emergence in psychology.
Why affordances?
The current state of psychology can be historically traced all the way back to the first attempts at studying visual perception, where scientists tried to explain how we came to see the world as it is. The theory that survived was that of the retina image, where light rays emanating from a source would be reflected upon an object before entering our eyes, thereby casting an image of the object onto the back of our retinas, just like how an image would be projected upon an opaque projector screen. There was a huge problem, though – the physics of optics and physiology of our lenses meant that this image was flipped vertically and horizontally, and also two-dimensional. On the contrary, our experience of the world was that of upright and three-dimensional objects. How do we go from images to objects?
To solve this, early scientists suggested that there was something in our minds that recognised this grossly ambiguous image, and through a series of computational steps, fixed the retina image to give it its experienced meaning. This ‘something’ was referred to as the central executive, and it was not without problems. For instance, how did the central executive know that the image was wrong and needed to be fixed? Why didn’t it just assume that this was how the world really was? And even after this, how did the central executive know how to fix the meaningless image? Scientists like Helmholtz argued that the central executive utilised past experiences of the world as a template to fix these vague inputs, but then we would ask, where did these past experiences come from? Given that current traditional approaches to psychology all stem from this Helmholtzian program, they incur what is known as a loan of intelligence that can’t be repaid – we simply can’t explain where and how this initial knowledge (to guide the fixing of subsequent sensations) came to be.
In general terms, the above approach assumes that what is sensed is inherently meaningless, and to which meaning is added through a process of enrichment by the central executive. The many issues of this account led Gibson to assume that meaning is already inherent in the object of perception (i.e., the environment!). This means that perception changes from a two-step process (of sensation and then enrichment) to a one-step process where organisms are simply required to detect the meaning-laden properties of the environment. Given another of Gibson’s assumptions that the primary goal of perception is for action, we take these meaning-laden properties to be related to possibilities for action, or otherwise, affordances!
What are affordances: dispositions or relations?
So, we’ve established the need for assuming affordances exist. But in what way do they exist, exactly? There are at least two accounts on the ontology of affordances. Turvey (1992) was the first to formalise affordances as dispositional properties of the environment that allowed for action to be performed. The complementary action-relevant dispositional property of the organism was termed effectivities (i.e., the abilities and size of the organism). Together, the affordance and complementary effectivity combined to result in the actualisation of the behaviour. For example, according to this account, the pavement has the affordance of being walk-on-able (in that it has an intrinsic property that grants it the possibility of being walked on), while an adult human might have the effectivity of walking (it has the dispositional property that grants it the ability to walk). Put together, we get a system with a relational property of adult-human-walking-on-pavement, that is, the manifestation of the actual walking behaviour! To the extent that this is a relational property is debatable – typically within philosophy, events and occurrences aren’t construed to be properties, so this ‘property’ might very well just be an occurrence that emerges from the adult-pavement system.
On the other side, there are those who instead view affordances as existing by virtue of the relation between animal and environment. For example, Stoffregen (2003) argues that affordances are relational properties that emerge from the animal-environment system, whereas Chemero (2003, 2009) posits that affordances aren’t even properties, but just relations within this system. At least in Stoffregen’s case, the idea is that the environment possesses some physical property, while the animal possesses some ability-related property. Combining these two properties then gives rise to a higher-order property that is the affordance. Using the example from above, the affordance of walk-on-ability is a result of the interaction between (say) the rigidity of the pavement and the adult human’s ability to walk. The given behaviour then necessarily occurs when the affordance co-occurs with the appropriate intentionality.
For me, the above analysis doesn’t make much sense. I’m still thinking and learning about this, so I can’t pretend that I can articulate it formally, but a reading of Mumford (1998) on dispositions defends what he calls a disposition-categorical distinction. Roughly, he argues that both dispositional and categorical properties are both categorical with a dispositional nature. The only way they differ is the kind of dispositional nature involved. More relevantly here, different categorical properties can interact to form a higher-order dispositional property. For instance, sugar has the dispositional property of solubility, and the categorical property of having a certain type of lattice structure. This categorical property (combined with other categorical properties) of sugar is what gives it the dispositional property of solubility. Reflected upon Stoffregen’s analysis, there is simply no need to confer the affordance property onto the animal-environment system. The environment properties he proposes might simply be analogous to Mumford’s categorical properties, a combination of which gives rise to a dispositional affordance property that still belongs to the environment, not the system.
I’m not as familiar with Chemero’s (or even Rietveld’s) affordances-as-relations analyses yet, but I suspect they might suffer from similar issues. It seems like the affordance-as-relations argument largely came about from a) the vagueness of Gibson’s initial definition of affordances and b) a general discontent with the affordance-as-dispositions analysis. Briefly, I point out one alleged issue with the dispositional account:
- The issue: Dispositions entail compulsory manifestation in the presence of the effectivity, yet unrealised affordances exist (Stoffregen)
- The reply: This analysis can still work if we’re willing to accept organisms as high-dimensional beings where a high bar of effectivity (which might include intentionality, attention, motivation etc) is required to be reached for behaviour to necessarily manifest. Also, a preliminary reading of Mumford suggests he might be reluctant to define dispositions with respect to their ‘necessary’ behavioural manifestations. Perhaps there is room to update Turvey’s initial formalism of affordances with insights from the developing philosophy of dispositions.
And even more briefly, the affordance-as-relations account is not without issues either. It doesn’t account for how affordances can exist in the absence of the animal, or how individuals learn to perceive the affordances of actions that they can’t currently effect. To this end, I’m going to let this writing piece be a reminder that these are areas that I’m not very confident of articulating yet. At the moment, I’m quite convinced that affordances are dispositions, not relations. However, I’m not too sure how the above two issues are necessarily damning critiques of the relations account. Chemero does try to solve at least the first issue, and I’ll have to take a closer look at whether it is satisfactory.
Andrew’s research programme
With regard to the second issue, this is something that I think might benefit from a better understanding of Andrew’s graphical operationalisation of affordances as the set of all combinations of the action-relevant parameters that result in hits (or misses) (Wilson, 2025). One insight from this analysis is that there is a distinction between perceiving the affordance (the whole area of the graph), and perceiving which parts of the graph one can currently effect (a smaller area of the graph based on one’s effectivities).
This might allow us to categorise the different stages of skill development. On one end of the spectrum, complete beginners (e.g., for throwing), might not be able to a) perceive the affordance of throwing (the whole graph; i.e., they can’t perceive the set of possible parameter combinations that will result in successful and unsuccessful throws) and b) perceive the actions that are actualisable given their abilities (you can’t intentionally effect behaviours if you can’t perceive the affordances; the latter is a necessary condition for the former). As they get more skilled, they may a) start to perceive more of the affordance, but b) still lack the effectivities to manifest the behaviour (might this suggest that perceptual skills develop faster than motor skills?). Still, the perception of the expanded affordance space acts as a target to which the intermediate thrower can now direct their efforts in order to further develop their throwing. Finally, a skilled individual might be able to perceive the whole affordance space, and is able to effect even more of the possible behaviours that live in this space.
Here, one question is whether the whole affordance space could ever be perceived – while the affordance space is theoretically infinite, surely there is a limit at which we can no longer perceive that part of the space? If this is so, and the affordance is defined as the entire graphical space, would it not be inaccurate to claim that one perceives an affordance? Would it be more accurate to say that one perceives more or less of an affordance space? Wilson et al. (2019) define affordances as the perceptible, action-relevant subset of the dispositional properties of a task. The above question casts doubt on the ‘perceptible’ aspect of the definition. This also reminds me of Gibson’s quote about how the central question isn’t whether affordances exist or not, but whether there is information about them that is detectable. This begs the question, are there affordances that exist that we cannot (yet, but also never ever) perceive?
Also, another question is as follows: Comparing a complete beginner to an MLB pitcher coming back from injury. Assume that at one time point, their ability to actualise the behaviour of targeted throwing is the same – both can’t accurately hit a 5m target. Yet, according to the above analysis, the MLB player should have access to the whole affordance space, that is, they can accurately perceive all the possible release parameter combinations that lead to hits and misses, whereas this ability is lacking for the beginner. How might we experimentally demonstrate this difference? Can we assume that, if asked to make yes/no perceptual judgements of whether or not they think they can hit a 5m target, that both individuals would say no? And if so, what else can be done to elucidate this difference in access to the affordance space? Given that the perceptual judgement paradigm is key to Warren's analysis of affordance as pi numbers (e.g., Warren & Whang, 1987), this case might point to a weakness of Warren’s method – while it might correctly identify that both the injured MLB player and beginner perceive that they can’t actualise the throwing behaviour (the graph scaled to their action capabilities), it says nothing about their ability to perceive the affordance (the entire graph). And in fact, it will likely lead (and probably already has led) to the false conclusion that both individuals are perceiving the same affordance, or that their affordance perceptual performance is the same.
How do affordances guide behaviour?
This final section will touch on one more aspect of affordances. While we focused on the ontological nature of affordances previously, this section is more concerned with the role affordances play in perception and action. A more detailed analysis will come when I write about information. But very briefly, affordances can be characterised as the action-relevant dynamic properties of a task that uniquely create kinematic information in the energy media that in turn specifies or map 1:1 to those affordances. This specifying information acts as the contact between organism and environment. And finally, another question is whether affordances are for action selection or for action control. Warren has it that affordances help select the action mode, before information is used for the control of that action. Meanwhile, Fajen (2007) argues that affordances are baked into action control through calibrating the information detected on an action-relevant scale. I will make clear how I think this happens when I write about information.
Gaps and future steps
Andrew’s research programme
With regard to the second issue, this is something that I think might benefit from a better understanding of Andrew’s graphical operationalisation of affordances as the set of all combinations of the action-relevant parameters that result in hits (or misses) (Wilson, 2025). One insight from this analysis is that there is a distinction between perceiving the affordance (the whole area of the graph), and perceiving which parts of the graph one can currently effect (a smaller area of the graph based on one’s effectivities).
This might allow us to categorise the different stages of skill development. On one end of the spectrum, complete beginners (e.g., for throwing), might not be able to a) perceive the affordance of throwing (the whole graph; i.e., they can’t perceive the set of possible parameter combinations that will result in successful and unsuccessful throws) and b) perceive the actions that are actualisable given their abilities (you can’t intentionally effect behaviours if you can’t perceive the affordances; the latter is a necessary condition for the former). As they get more skilled, they may a) start to perceive more of the affordance, but b) still lack the effectivities to manifest the behaviour (might this suggest that perceptual skills develop faster than motor skills?). Still, the perception of the expanded affordance space acts as a target to which the intermediate thrower can now direct their efforts in order to further develop their throwing. Finally, a skilled individual might be able to perceive the whole affordance space, and is able to effect even more of the possible behaviours that live in this space.
Here, one question is whether the whole affordance space could ever be perceived – while the affordance space is theoretically infinite, surely there is a limit at which we can no longer perceive that part of the space? If this is so, and the affordance is defined as the entire graphical space, would it not be inaccurate to claim that one perceives an affordance? Would it be more accurate to say that one perceives more or less of an affordance space? Wilson et al. (2019) define affordances as the perceptible, action-relevant subset of the dispositional properties of a task. The above question casts doubt on the ‘perceptible’ aspect of the definition. This also reminds me of Gibson’s quote about how the central question isn’t whether affordances exist or not, but whether there is information about them that is detectable. This begs the question, are there affordances that exist that we cannot (yet, but also never ever) perceive?
Also, another question is as follows: Comparing a complete beginner to an MLB pitcher coming back from injury. Assume that at one time point, their ability to actualise the behaviour of targeted throwing is the same – both can’t accurately hit a 5m target. Yet, according to the above analysis, the MLB player should have access to the whole affordance space, that is, they can accurately perceive all the possible release parameter combinations that lead to hits and misses, whereas this ability is lacking for the beginner. How might we experimentally demonstrate this difference? Can we assume that, if asked to make yes/no perceptual judgements of whether or not they think they can hit a 5m target, that both individuals would say no? And if so, what else can be done to elucidate this difference in access to the affordance space? Given that the perceptual judgement paradigm is key to Warren's analysis of affordance as pi numbers (e.g., Warren & Whang, 1987), this case might point to a weakness of Warren’s method – while it might correctly identify that both the injured MLB player and beginner perceive that they can’t actualise the throwing behaviour (the graph scaled to their action capabilities), it says nothing about their ability to perceive the affordance (the entire graph). And in fact, it will likely lead (and probably already has led) to the false conclusion that both individuals are perceiving the same affordance, or that their affordance perceptual performance is the same.
How do affordances guide behaviour?
This final section will touch on one more aspect of affordances. While we focused on the ontological nature of affordances previously, this section is more concerned with the role affordances play in perception and action. A more detailed analysis will come when I write about information. But very briefly, affordances can be characterised as the action-relevant dynamic properties of a task that uniquely create kinematic information in the energy media that in turn specifies or map 1:1 to those affordances. This specifying information acts as the contact between organism and environment. And finally, another question is whether affordances are for action selection or for action control. Warren has it that affordances help select the action mode, before information is used for the control of that action. Meanwhile, Fajen (2007) argues that affordances are baked into action control through calibrating the information detected on an action-relevant scale. I will make clear how I think this happens when I write about information.
Gaps and future steps
Affordances are real dispositional properties inherent to the environment. Their existence depends on neither the presence of an organism (as relational accounts would have it) nor the existence of a complementary effectivity (as Turvey would have it). There are still some questions that I have with regards to 1) proposed issues to both the relations and dispositional accounts, 2) studying affordances with Andrew’s research programme, and 3) how calibration and affordance-based control work. To this end, the short-term goals are to 1) read Mumford (1998) and spend more time in the relations papers, 2) read all the other throwing-related papers, and 3) read Fajen's empirical work. The absolutely ambitious and lofty long-term goal is to contact Mumford and actually write this updated dispositional account of affordances supported by both theoretical advancements in the philosophy of dispositions and empirical work from Andrew’s program :)
References
Chemero, A. (2003). An outline of a theory of affordances. Ecological Psychology, 15(2), 181–195. https://doi.org/10.1207/s15326969eco1502_5
Chemero, A. (2009). Radical embodied cognitive Science. MIT Press.
Fajen, B. R. (2007). Affordance-Based control of visually guided action. Ecological Psychology, 19(4), 383–410. https://doi.org/10.1080/10407410701557877
Mumford, S. (1998). Dispositions. Clarendon Press.
Stoffregen, T. A. (2003). Affordances as properties of the Animal-Environment System. Ecological Psychology, 15(2), 115–134. https://doi.org/10.1207/s15326969eco1502_2
Turvey, M. (1992). Affordances and Prospective Control: An outline of the Ontology. Ecological Psychology, 4(3), 173–187. https://doi.org/10.1207/s15326969eco0403_3
Warren, W. H., & Whang, S. (1987). Visual guidance of walking through apertures: Body-scaled information for affordances. Journal of Experimental Psychology Human Perception & Performance, 13(3), 371–383. https://doi.org/10.1037/0096-1523.13.3.371
Wilson, A. D., Zhu, Q., & Bingham, G. P. (2019). Affordances and the ecological approach to throwing for long distances and accuracy. In The MIT Press eBooks (pp. 581–600). https://doi.org/10.7551/mitpress/10764.003.0032
No comments:
Post a Comment