Wednesday, 4 June 2014

Affordances are not probabilistic functions

The journal Ecological Psychology is hosting a special issue with papers from a Festschrift for Herb Pick. Karen Adolph and John Franchak have a paper that caught my eye about treating affordances as probabilistic functions, effectively applying standard psychophysical techniques to the study of affordance perception. 

The idea is this: affordance research typically treats affordances as all-or-none, categorical properties. You can either reach that object or you can't; you can either pass through that aperture without turning or you can't. You then measure a bunch of people doing the task as you alter some key parameter (e.g. the distance to the target, or the width of the aperture) and find the critical point, the value of some body-scaled measurement of the parameter where behaviour switches from success to failure. For instance, you might express the aperture width in terms of the shoulder width and look for the common value of this ratio where people switch their behaviour from turning to not turning.

Franchak & Adolph think this is misleading. Instead of critical points, they suggest it is more common to find thresholds with some level of variability: 
Figure 1. Categorical vs probabilistic functions
As an example, they measured themselves walking through an aperture and plotted the data:
Figure 2. Data showing performance as the authors try to walk through an aperture without turning
They call the function they fit an affordance function, and there are parameters you can extract including the mean (where the function crosses 50%) and the variability (the slope of the line between 0% and 100% success). 

This is all standard psychophysics. There are a variety of ways of fitting these functions and modern adaptive techniques that compute the next width to present so as to minimise the number of trials required to get a reliable estimate of the threshold. If you are interested, there is a recent excellent textbook by Kingdom & Prins (2010) that will teach you everything you need to know about doing this work. I actually like the idea of using adaptive techniques to drive the parameter selection, as these methods are typically very efficient.

But the resulting plot is not the affordance. 

These functions are not affordances
Affordances are dispositional properties of the world. Psychophysics does not measure properties of the world. It measures the psychological response to those properties, and finds a function that relates the property to the response. Franchak and Adolph mix these levels of analysis up, and argue that the function reveals that affordances can be probabilistic. What it actually shows is that people effect affordances probabilistically. Perception of that affordance, as measured by performance, is variable. 

This is well known by psychophysicists, by the way; they would never claim that the property of the world is described by the function relating that property to the perceptual response to that property. To do so is simply a category error.

 I think this case reflects the general confusion surrounding affordances - are they dispositions, are they relations? (They are dispositionsproperties of the environment picked out by properties of the organism.) People seem to really struggle with this idea, but it's important to get right or else we will be looking in the wrong place for our answers.

This is still useful though
Affordances are not probabilistic. However, the authors are not wrong when they identify that a given aperture close to threshold sometimes allows you through and sometimes doesn't. There is variability in performance. The effecting of affordances is probabilistic.

They discuss (quite correctly) that the static geometry of the body (things like shoulder width, or leg length) is not the relevant metric for the perception of affordances. Chemero discusses this in his book (Chapter 7, reviewed here and here); Proffitt talks about 'effort' based metrics to try and address this (e.g. here). Numbers relating to things like shoulder width were a useful proxy for early affordance research (e.g. the classic Warren, 1984 and Warren & Whang, 1987) because they helped show that people do not perceive the world in the units of physics, but in terms of their ability to perform the task - in other words, that they were indeed perceiving the affordances. But there is more to it than this.

More recent work has shown that a better measurement of the person's ability to effect the affordance is in the dynamics of the execution of the action (e.g. Snapp-Childs & Bingham, 2009 plus the references cited in Franchak & Adolph). What matters is how the action unfolds over time; is it a highly variable action? Is that variability noise or something systematic, like postural sway? Both noise and sway affect your ability to walk through an aperture in ways you can perceive and build into your response; if you sway a lot, for example, you need a bigger aperture than someone the same size but who sways less than you. Shoulder width is therefore not the key, but the motion of that width through the aperture is (see also Mon-Williams & Bingham, 2008 for a related analysis of prehension).

What's important about this is that the dynamics of an action can vary from trial to trial in a way that your body geometry doesn't. You might sway more if you are tired, for example. This psychophysical analysis is perfect for identifying this sort of variation. By definition, the function maps a physical parameter onto an individual's psychological response to that physical parameter. Variability in how an individual effects an affordance because of the dynamics of that effecting can be quantified and analysed using this technique. So that's nice, I think; but that's about the limit.

An important note about units
Franchak and Adolph note the following at the end:
In this paper, we have exclusively used extrinsic units to describe affordances—centimeters to describe affordances for passing through openings and degrees to describe affordances for walking down slanting surfaces. Although we agree that intrinsic units are important for understanding the specifying information for perception, researchers’ choice to only measure intrinsic units assumes rather than identifies the factors that determine a particular affordance. Prior to  gathering evidence about putative intrinsic units, a more agnostic, empirical approach is to measure affordances in extrinsic units.
The issue of identifying the action related units in which the affordance lives is indeed a tricky one. However, this agnostic approach is not the solution; task dynamics is. A detailed task analysis can provide a dynamical level description of the task at hand, which then provides a constrained list of task relevant variables (resources) and the information for those variables. This (I'm arguing in an upcoming paper on throwing) is the future of affordances; more on that later when that paper is actually done. 

Affordances are not probabilistic and they should not be equated to the psychophysical function relating the world and behaviour. These techniques might be a handy way to look at affordance data in order to start getting serious about individual variation and the factors that drive that variation, but that's about it. The future of affordance research is task dynamics, not psychophysics, and this paper simply spends too much time confusing various levels of analysis. 

Franchak, J. M., & Adolph, K. E. (2014). Affordances as probabilistic functions: Implications for development, perceptio nand decisions for action. Ecological Psychology, 26 (1-2),109-124. Download

Kingdom, F. A. A., & Prins, N. (2010). Psychophysics: A practical introduction. New York: Academic Press.

Mon-Williams, M. & Bingham, G.P. (2011). Discovering affordances that determine the spatial structure of reach-to-grasp movements. Experimental Brain Research, 211(1), 145-160. Download

Snapp-Childs, W. & Bingham, G.P. (2009). The affordance of barrier crossing in young children exhibits dynamic, not geometric, similarity. Experimental Brain Research, 198(4), 527-533. Download

Warren, W. H. (1984). Perceiving affordances: Visual guidance of  stair climbing. Journal of Experimental Psychology: Human Perception and Performance, 10, 683-703.

Warren, W. H., & Whang, S. (1987). Visual guidance of walking through apertures: Body-scaled information for affordances. Journal of Experimental Psychology: Human Perception and Performance, 13, 371-383.


  1. Correct me if I'm wrong, but this seems to resolve a confusion that troubled me in our previous exchanges about affordances. Eg, does a ladder sixty feet away offer the affordance "climbable". If - as I recall - immediate effectability is part of the definition of "affordance", then the answer is clearly "no". But if an affordance is a context-independent dispositional property of the world, then the answer seems clearly "yes".

    As you describe such situations here, the resolution seems to be that the focus should be on the possibility of at least one context in which a hypothesized affordability could be immediately effected. Then an affordance is always present, but in any specific context may not be immediately effectable.

    Digging way into the past, this seems to resolve the issue of Ken Aizawa's exploding box. I may be projecting my ignorance onto others, but as best I recall we all assumed at the time that the box did offer the affordance "pick-up-able", but it could never be effected because the box would explode due to any contact (or perhaps even mere proximity of a subject). But if the possibility of immediate effectability in at least one context is part of the definition of "affordance", then the box doesn't offer "pick-up-able" since there is no possibility of that hypothesized affordance being effected in any (straightforward) context.

    1. Then an affordance is always present, but in any specific context may not be immediately effectable.
      This is sort of true. Try this:

      The affordance is a property of the world that is picked out by properties of the organism engaged in a task. It is always definable (from a third person analysis) but it is not always available (from the first person perspective) if I am in the same space as the affordance but engaged in a different task, or if the affordance is not currently specified for some reason, or if i fail to detect the information (if I haven't learned it, or if I fail to attend to it).

      The ladder is therefore always climbable by me, and it always offers that affordance to me (the third person view) but I may not take up the offer because I'm not looking to climb right now and I'm attending to different things, or because I miss the ladder (the first person view).

      The box is always perceived as being pick-up-able because the only properties that are creating information are ones that specify that affordance. The exploding part doesn't create information ahead of time and is therefore irrelevant to the perceiver (until it explodes, in which case both information and relevance show up in a rush :). Ken's problem is he doesn't understand information.

    2. That all seems internally consistent, but it means that immediate effectability is NOT part of the definition of "affordance". A distant ladder's affordance "climbable" is not immediately effectable, although it may be in a different context. The box's affordance "pick-up-able" is never effectable in fact, although I get your point about information. Perhaps there needs to be a qualification something like "an affordance offered to a subject must appear to be immediately effectable in some context as determined by the subject from the available information".

    3. The immediate effectibilty thing is, as I recall, a side issue from the dispositional account. If affordances are dispositions then their effecting is mandatory in the presence of the complementary properties (in the organism). We don't run around effecting everything, therefore affordances aren't dispositions (say Chemero and others).

      The actual solution is that I am not always in a state that provides the complementary properties to every affordance in my environment. Stairs still have the disposition to support my climbing them even when I am currently sitting down (they have to in order to help explain why I get up and start climbing) but I am not currently able to effect this action.

      I've come to think that this is not complicated but it's made so by defining affordances as 'opportunities for behaviour'. This is sufficiently imprecise that it creates straw men to attack, I think.

      So being immediately effectible is not part of the definition of affordances.

    4. Not commenting on your post (because I simply agree), and instead commenting on the comments:

      Charles, this is indeed a not-fully-worked-out problem. The path Andrew wants to walk down looks nice at the start, but it leads to a most extreme behaviorism, and, based on other conversations, I don't think he wants to end up there. What to do?

      Gibson cheated a bit. Whereas much of biology cheats by assuming "the environment", Gibson assumed "the organism". But is the "you" who is 60 feet from the ladder, sitting at a table, eating pie, the same you who in half an hour will have one hand on the ladder and the other clutching a roofing hammer? Andrew's comments are aimed squarely at that cheat. Again, what to do?

      I suspect that the way to get a little more play is to make a weaker version of Andrew's move. This would mean saying something like: "An affordance exists when organism X could do behavior Y, given conditions Z." We then need certain things to be allowed in Z, and others not. To phrase it way too crudely "If it wanted to" and "If the ladder was closer" are allowed, but "If it had twice as much muscle" or "If it magically grew a third leg" are not. I am not quite sure how to get that technically rigorous, but I am pretty sure it is possible.

  2. What do you think about equating social communication signals as affordances? For example in a territorial confrontation of two organisms, if one were to display aggression would this display be considered an affordance that the other organism acts upon (either to display further aggression or submission). In this case the act, if it could be considered an affordance would be variable as well as the perception and the effect based on that act in the observing organism.

    1. At this point I've chosen to reserve the word 'affordance' to describe properties of the environment that support the perceptual control of actions. This is the technical meaning of the term.

      Now, there are clearly perceivable properties of social situations/events/tasks that create opportunities for action. Are these affordances? Maybe not; they will be like affordances, in that they create information and are about behaviour, but it might be the case that not everything we directly perceive can sensibly be called an affordance.

      To figure this out will require working out Sabrina's information taxonomy in more detail and then figuring out what kind of information is present in social situations and what that information is about.

    2. Eric, the problem is that you can usually do communicatory signals, e.g., I am almost always afforded the ability to do one side of a conversation. The other person speaking does nothing to create the opportunity for me to speak.

      The way to get a foot in the door for communication (which has been worked out by Nick Thompson, and by Don Owings) is to talk about what I can accomplish through influencing another (or, how the structure of my actions can create structure in others' actions, following law-like principles) . Perhaps a certain set of circumstances afford my getting a steak delivered to the location I am sitting at by means of producing the right sounds. That I can make sounds while sitting at the table is not very interesting, but it is interesting that I can make a steak arrive.

      I suspect that this can be worked out so that it is almost identical to standard affordance talk, but it is a bad place to start. Best to get straightforward affordances down pat first. Then, best to focus on communication, broadly conceived, including comparative work. Stay away from language as long as possible, because it is misleading. Also, take Skinner's work on communication seriously. It is not perfect, but it is very insightful, an some of it must be either adopted or reinvented.

  3. Dear all apologies for being ignorant but I was still a bit unsure about the differences between dispositional vs relational affordances.

    I understand the idea you (andrew) noted in an earlier post that objects as salt are disposed to dissolve in water because of the chemical properties of ionic salts and the electrical charges of water molecules (qouted). But taking a different example, a ladder only affords 'climable' when the first person (ever) has decided to climb it. Is the affordance still a dispositional quality if it's existence depends on the choice of the agent to act on it?
    I guess this is the case but what if I wouldn't know that I could climb the ladder and the affordance is still dispositional (for the first person) to the ladder but would not afford this response for me. Does my inexperience to act on the affordance similar to the first person relate to effectibillity or it this something else?

    1. Eric covered most of this, but

      But taking a different example, a ladder only affords 'climable' when the first person (ever) has decided to climb it. Is the affordance still a dispositional quality if it's existence depends on the choice of the agent to act on it?
      Your premise is false, basically.

      Dispositions are real properties of objects that complement real dispositional properties of other objects in such a way that when these things come together, the disposition occurs. Affordances are these kind of properties, specifically ones that are complemented by organisms.

      Now, you could always pick out a dispositional property of an object and say 'well if there were such things as 15 legged snails about 1m long that property would totally afford that organism 'gerbling'.' If you can adequately define the two complimentary things you can define the disposition. However, this is ecological psychology and so we are only interested in things that actually exist in the ecologies of actual organisms. We are therefore into 'climbability' and not 'gerbling'.

      But here's the thing. Dispositions can be defined independently of the presence of anything that can effect them. They can exist prior to their complement, they just don't do anything until the complement shows up. So the ladder always affords climbing to an organism in the right, complementary state (see Eric's description). This is important; they have to exist independently of the organism for them to be the kinds of things we can learn to use.

      Whether or not you perceive this affordance is a whole other matter. Does the property structure an energy array? Is that structure above threshold? Has the organism spent time trying to engage in the task before, thus making the structure available to be learned? and so on.

      Always remember; affordances are not even half of the ecological analysis. They are important but the real work is done in information about affordances.

      Not sure if that helps, but anyway... :)

    2. Yes, this makes it much more clear! Independent of actors, objects afford all possible actions given that they structure an energy array, it depends on the information people use to act on the disposition of the object. It also makes sense if I try to think of the first person ever to have climbed a ladder. The property of climbing is already present in the ladder and independent of the concept of climbing. Even if we provide language (climbing) and motor experience (stepping on ladder bars) to the affordance the existence of the affordance does not depend on it.

  4. Tim,
    To make the dispositional model work, you need to stipulate both properties of the object an the organism. If the ladder has all the properties necessary to be climbed AND the organism is the type of organism that will climb given the opportunity, then, when the two meet, the disposition obtains and there is an organism at the top of the ladder.

    In this way of thinking, your having-had-proper-experience, having-the-desire, having-the-caloric-power, etc., etc., all combine on philosophically-equal footing to make you the type of organism that will climb when you encounter the ladder.

    So, to answer your question, if you can manage to get rid of everything dualistic that is typically implied by the word "choice", then: Yes, one could say that the existence of the affordance is dependent upon that.

  5. P.S. Andrew and Sabrina wrote a great section on this for the chapter we have forthcoming. I will leave it up to him to excerpt if he wants to.

  6. OK, based on these comments and the chapter, I think I finally (after “only” two years!) understand how the information view works. Objects in an environment are, in general, emitting structured energy. Such a structure may contain information which can be viewed as being an invitation to certain organisms to enter into a dynamic interaction with the object. Call such an invitation an "affordance" and the interaction a "task". In order for the organism to accept the affordance, it must be capable of detecting the information and be disposed to immediately initiate the task.


    This way of describing the situation avoids the "immediate effecting" problem of the distal ladder by viewing the whole process - from the moment at which the affordance offered by the ladder is first detected and accepted to the moment at which the ladder has been climbed - as a single task which is immediately initiated but takes time to execute.

    I view speech as involving a speaker - an "object" in the environment - who emits structured aural energy (and possibly energy in other modes) thereby offering an affordance to a hearer - an organism in the environment - to activate an existing behavioral disposition (or in some instances to create a new disposition). A hearer who has learned to extract the information content of the affordance may - depending on context - be disposed to accept the invitation and to reply. A simple conversation then can be thought of as an exchange of offered and accepted affordances and completed tasks (ie, replies). In this view, speech seems to fit right into the above description.

  7. Charles,
    I would tweak the language a bit, but my overall answer would be "yes", that sounds right.

    Communication complicates the story for many reasons, but the hope would be that it can, "fit right in".

    1. Just to drop more hints about Sabrina's work; language won't 'fit right in' and the lack of progress in the field is partly a function of trying to talk about language in terms of specification and affordances (two things which are actually solely about the online perceptual control of behaviour). Her taxonomy is about the fact that language isn't about affordances and that information comes in a variety of flavours but that this is a) ok b) useful and c) still ecological.

      If you want more, you'll have to hound her to get working on the post for here based on her UConn talk :)

  8. Hi Eric -

    Since I think getting the language right is crucial for fruitful discussion, how about suggesting the tweaks.

    Also, since my interest is specifically the comm, I'm interested in what complications you envision. I certainly see that in general some environments, affordances, and dispositions will be much more complex than others, but I see nothing peculiar to comm in particular. The affordance embedded in the aural energy manifest in a speaker's vocalizing "Please pass the salt" and a hearer's acceptance of the affordance and activation of learned dispositions to vocalize "Certainly" and to pass the salt seem relatively straightforward to me.

  9. To put a little meat on that last claim, I note that when confronted with the affordance "What do you think about X?", I often have dispositions to pontificate on X with some level of credibility based on some fairly limited exposure to relevant material (eg, when X="affordances"!). OTOH, after thirty years of opportunities to learn to either execute or return a top-spin tennis serve, I was never able to develop the relevant dispositions to respond with even minimal competence. So, for me it appears that dispositions to execute even a relatively esoteric BSing task often come much more easily than dispositions to execute more "purely physical" tasks.

  10. I don't want to nit pick too much, but: hardly anything "emits" light, most things just reflect light, and thereby add structure to ambient energy. Some aspects of that structured light can specify the opportunity to act (rather that "invite"). That is, those structures are specific to situations in which the opportunity exists. (Andrew would also add, I think, that there is information available to specify who can take advantage of the affordance, but I am ambivalent about that.)

    As for the complications with communication:
    The BIG complication is that it is unclear how specification works in that context. One could imagine that the salt shaker shapes light in a way that specifies the pick-up-able-ness of the shaker, so that the right type of organism could "resonate" with that information and accurately perform the action (picking it up). But what about the other person at the table specifies that I could move my mouthflap in the right way while exhaling so as to get them to pass me the salt? We all agree that the affordance is there, but can we talk about my perceiving the affordance in the same way I perceive that I can pick up the shaker?

    We might well be able to get that way of thinking to work, but it is certainly not the most natural context for talk of specification. This stuff might be relevant, I'm not entirely sure:

    1. As for the complications with communication:
      The BIG complication is that it is unclear how specification works in that context.

      It probably won't. Sabrina was just at UConn talking about this and is working on a post summarising the talk, plus there will be a conference proceedings paper. It's awesome and will blow your minds :)

      The emit/reflect thing isn't nitpicking, by the by. Emitted light is typically not structured, and the kind of structure required to control behaviour only comes about because of the ecological laws governing reflection.

      Andrew would also add, I think, that there is information available to specify who can take advantage of the affordance
      I'm not sure I would say this, or why I would.

  11. Well, I can't imagine that Charles actually meant "emit" (as we all know that most things don't emit light), so I assume he meant "reflect".

    As for the other, perhaps I did not phrase correctly. You were, a few years ago, really big on the idea that the light contained information on both what was up with the object and what was up with me, such that combined it specified that the body that was me could do the thing with that object. i.e., the light specifies something about the width of the doorway and the width of my shoulders, such that I can pass through without turning. I'm not quite sure how to say that better, but I haven't seen an indication that you changed your mind about this. Any chance you know what I'm talking about here better than I do and can help me out?

    1. Oh right. Well it might not just be light, but there has to be information about (eg) my shoulder width** or else information about the door can't get scaled correctly. This is calibration, and it's using information to scale other information (so it's not quite the same as the regular use of information to control behaviour).

      **Static body geometry is not where it's at, as I mention above but for argument's sake let's do it this way.

  12. Yes! I meant that thing you just said. Many thanks!

    I am not sold because of my background in development and learning theory. You can get darn accurate just by calibration to the specifying information about the external world (assuming your body is relatively stable in size, strength, etc.), without needing access to specifying information about yourself. Of course, it would help on-the-fly adjustments tremendously if the latter type of information was available. I'm just not sure it is necessary.

    I brought it up here because it is also an additional complication for the language story. What information specifies that I, myself, am capable of talking loud enough to be heard in this room? Certainly people are sometimes wrong about that (e.g., when you first find out you have laryngitis or right after successful voice lessons have made you better able to project). Do we need the added headache of determining the information available to my senses that specifies those things about me, or can we get by with atunement through "plain Jane" learning mechanisms?

    1. You can get darn accurate just by calibration to the specifying information about the external world (assuming your body is relatively stable in size, strength, etc.), without needing access to specifying information about yourself.
      Now it's important that static body geometry (ie something you can just count on as stable) is not the right calibration ruler. If perception is scaled by the dynamics of the execution of the action as described in the post then you must have information about that for the same reason you need information about anything - because you can't 'know' it otherwise (where 'know' is in scare quotes for obvious reasons).

  13. I chose "emit" to generalize since"reflect" applies only to light energy. defines "emit" as "to send forth", which seems pretty general.

    I don't understand all that confidence that structure in speech is a problem. Perhaps it stems from an assumption that speech processing comprises multiple steps: recognize individual words, then parse the resulting word string for grammatical structure (which of course doesn't follow natural laws), then determine meaning, then formulate a response.

    I don't envision aural processing as being like that, at least not in relatively simple cases. Just as visual sensory stimulation ultimately results in patterns of neural activity (information-bearing??), so too aural sensory stimulation. And at least in relatively simple cases, those patterns - together with context-dependent responses - presumably are what get "learned" in the brain. Think of instances where in a given context you can answer a question even before it has been completed or where you finish someone's sentence for them. There clearly isn't much, if any, grammatical parsing going on if you don't even have a complete sentence to work with. [See note below]

    Anyway, my impression is that the ecological approach does not envision processing of visual stimulation as involving parsing the input into objects and then building a model of the environment to be "parsed" according to some set of rules. If not, why envision processing of aural stimulation as being like that?

    Eric's comparison between speech and passing the salt shaker assumes a certain assignment of the roles of energy source and energy sensor. I don't see a problem with viewing the scenario as involving a salt shaker offering an affordance something like "salt-shake-out-of-able" which kicks off the task do-what's-required-to-get-the-shaker-in-hand, the first step of which is to ask that it be passed (just as the first step in climbing the distal ladder is to move to where it is). But I'm addressing the role assignment in which the speaker is an object that does indeed emit (not reflect) aural energy and that energy's structure does specify "opportunities to act" - ie, affordances - for some hearers.

    BTW, I'm really not arguing that speech processing fits into the ecological approach (if for no other reason than I'm not competent to do so), just playing devil's advocate. It would be interesting to me if it did since that would offer some support for my view of speech processing, but otherwise I have no skin in that game.

    Note: google is quite good at this, at least for well known movies; try "who played the ti" [... n man in Wizard of Oz]; or "I'll have wh" [... at she's having], a line from When Harry Met Sally.

  14. Charles,
    I'm fine with most of what you say in the last post, I'm just not sure it is "Ecological Psychology" proper. I don't think Eco-Psych can handle everything we do with only a few basic principles, and neither did Gibson. However, I agree with Gibson (and Turvey, and Shaw, and lots of other people) that it is a good thing to push the basic principles as far as they can go, and only admit in other things when we can't go further.

    Of note, your account mixes the eco-psych approach with an operant chaining approach (e.g., response A creates the indication that it is time to do B, creates the indication that it is time to do C, etc., with no "law like relations" necessary). Operant chaining definitely happens, and it is a powerful explanatory tool. Also, both specification and chaining are almost certainly needed for a full story of how linguistic communication works.

    However, if we wanted a "pure" eco-psych approach, we would need to argue, I think, that the detectable properties of the salt shaker (or some larger aspect of the situation containing she shaker) specify something about "Please pass the salt" (and "Pasame la sal por favor", and "моля премине солта"). I don't think that game is going to work, but I would be thrilled to be proven wrong.

    I think we can do an awful lot of communication (broadly defined) within a "pure" ecological approach, and I am not sure when we would need to let other things in, but I think we eventually will. If I find the time I'll do a post over on my blog soon about communication.

  15. I'm not familiar with "operant chaining", but it sounds like a way of breaking a complex task into simpler tasks that are more amenable to analysis. If so, I don't mean to be doing that. Based on Andrew's posts, I've (rightly or wrongly) inferred that the fundamental entity of analysis is the "task", which is defined by an organism in an "environment" (which I assume includes the state of the organism; if that's wrong, substitute "context") which has detected something in the environment that offers it the opportunity (affordance) to execute a behavioral disposition, ie, perform a task. The task has an objective, a start, and a termination. Eg, in the fly ball scenario, the environment is the playing field, a fielder, and a ball in flight that offers the affordance "catchable" to that fielder (requiring that the ball's path and the fielder's state meet certain requirements). The task objective is to move in such a way as to catch the ball, the start is the moment at which the affordance is detected, and the termination is the moment at which the ball either is caught or hits the ground.

    The simple speech environment I'm assuming comprises a speaker, a hearer, and an utterance by the speaker. The utterance offers the hearer the affordance "respondable-to" in that the utterance is assumed to be in a shared language, audible by the hearer, and "understood" in the special sense that it activates a behavioral disposition in the hearer that results in the response intended by the speaker. The hearer's task starts at the moment the affordance has been detected and terminates at the moment a response has been executed.

    Perhaps a conversation can be viewed as a case of "operant chaining" in which each utterance-response pair is a link in the chain, but that's a degree of complexity greater than I'm trying to address here.

    we would need to argue, I think, that the detectable properties of ... some larger aspect of the situation containing the shaker) specify something about "Please pass the salt"

    I'm not yet comfortable with the term "specify", so I'm going to describe the situation without using that word. An English speaker in a state of requiring access to salt might perceive a shaker in reach of a dinner companion as offering the companion the affordance "passable" and perceive the companion as offering the speaker the affordance "requestable". The latter affordance might activate a disposition in the speaker to utter "Please pass the salt shaker". If that formulation meets your requirement that detectable properties of the environment "specify something about" the utterance, then I agree with the quoted claim.

  16. I can go along with all that.

    So then, the remaining question is: Can we get "information" into the story, understood in the Gibsonian sense as "ambient energy patterns specific to a state of the world".

    Affordances are crucial components of Eco-Psych stories, and I agree with your description in those terms. But then we need to ask whether those things can be perceived, in the narrow sense that Eco-Psych uses (of "resonating" with specifying invariants).

  17. You guys are probably going to want to go read Sabrina's new post :)