-------Original Post--------
My proximate motivation for starting this blog is that I have become an active commenter on a couple of blogs, and I both envy the authors and feel bad when I want to post a reply elaborate enough that it might look like thread hijacking. With that in mind, I want to rift for a bit off of Charles Wolverton's reply to my comments on Sabrina Golonka's post over at Notes from Two Scientific Psychologists. I was trying to cut the difference between Ecological and traditional Behaviorist ways of thinking, and Charles said:
In the absence of the tone, the lever doesn't necessarily invite purposeful action, ie, doesn't offer an affordance. But as Eric notes, the rat has learned to associate the tone with a purpose - getting access to something offering "ingestible" - so that the tone's sounding results in the lever's otherwise purposeless offer of "graspable" becoming purposeful, and hence an affordance.
This made me feel great, but also a little squeamish. It is great, because Charles clearly got my point, and phrased it better than I did. It made me squeamish because of the very last word: affordance.
The term 'affordance' was introduced by Gibson, and has since been adopted by many others and adapted for many purposes. One challenging topic as I work on the incipient Ecological Psychology textbook is what to do about how broadly or how narrowly to use the term affordance. My initial inclination is to use a very restrictive reading of the term. The most natural use of the term 'affordance' is to refer to a behavioral action made possible by a matching between the crude physical features of an organism and the nuances of the immediate environment. My hand is the rough size and shape that allows the picking up of a coffee cup, my frame allows me to fit through standard doorways, etc. The affordance is thus the relationship between me and the objects in question. The further we get from such straightforward relationships, the more nervous I am about using the term 'affordance'.
One reason I am nervous, is because of how we (ecological psychologists) want 'affordance' to relate to the other key terms in our system 'invariants', 'specificity', and 'perceptual systems'. We are particularly interested in affordances that can be perceived, that is, affordances that structure ambient energy (e.g. light) in a manner that specifies the relevant physical properties of the object. When such specification is present, a properly attuned perceptual system can accurately detect the properties of the world. This is important, because it provides a definition for perception, which allows us to distinguish the perceptual process from other (potential) psychological processes. When we are attuned to the proper higher-order invariants, the affordances themselves become our 'percepts', they are given directly in perception.
So, back to the rat and the lever. It is very easy, with the most natural use of the term 'affordance' to talk about the rat perceiving the lever as affording pressing. The physical structure of the rat and the physical structure of the lever take care of that perfectly fine, and everything about the lever being press-able by a rat-like organism is specified in the ambient light (or so I assert for the purpose of this example). However, I'm nervous saying that the affordance of the lever changes when the tone sounds. Yes, the rat normally lives in a world where pressing the lever is boring, but when the tone sounds the rat suddenly finds itself in a world where pressing the lever produces food. Yes, because the rat has jumped between those two worlds many times before, the rat is now so built that it presses the lever many times when the tone sounds, but rarely otherwise. Yes, one can objectively describe the rat's behavior towards the lever as food-directed when the tone is sounding, but no such description is appropriate the rest of the time. I'm a crazy behaviorist, I'm happy with all that. But I'm nervous saying that the 'affordance' of the lever has changed. The physical relationship between the rat and the lever has stayed the same, and the lever affords pressing equally in both worlds.
Am I being too picky?
I think maybe. I'd pick on Charles' use of 'association', etc, but the idea of bootstrapping yourself into higher order affordances seems unproblematic and actually quite important.
ReplyDeleteThe affordance of the lever and the affordance of the lever that produces food are indeed two different things, analogous to Sabrina's take on the neural correlates of these things in that post. The lever isn't simply filling a structuralist slot, but is participating in two different higher order event structures.
And while I disagree with Tony that affordances are relations, I do think he's right in that body scale isn't generally going to do it either.
Eric -
ReplyDeleteWell, that was a treat way beyond what you could possibly imagine! As you perhaps have inferred, I'm a rank amateur at this (retired satcom engineer) and after reflecting on my comments, decided they were OT and naive - at best. So, to find that not only were they thought to be worth a response at all but were even the motivator for a blog post was indeed a treat. Plus, being a committed pragmatist (ala Rorty et al) I'm especially eager to read further posts here.
As for the issue of "affordance", I fully appreciate the problems with the word (the exchange on Ken's blog that I mentioned before was about that issue and went on for months - literally). Note that in my comment I explicitly redefined "affordance" by adding "purposeful", an attempt to resolve the issue of what causes a subject to choose from among the many affordances generally available at any time for actualization. There is no doubt a better (in the sense of being more consistent with traditional usage) way of capturing the general idea, but not being very familiar with traditional usage, adding "purpose" is the best I can do.
I look at the sounding of the tone as a change in the "nuances of the immediate environment", so that my way of thinking about the scenario seems consistent with your "most natural use" of "affordance". But although I am a stickler for a consistent vocabulary (in the "vocabulary" vocabulary sense), I care little about what comprises any particular one and am happy to redefine or drop any specific word in any specific vocabulary.
(Clearly, to say that the lever doesn't offer an affordance before the tone sounds was wrong - thanks for politely not pointing that out.)
As for being too picky, I'll best you every time. Eg:
We are particularly interested in affordances that can be perceived, that is, affordances that structure ambient energy (e.g. light) in a manner that specifies the relevant physical properties of the object. When such specification is present, a properly attuned perceptual system can accurately detect the properties of the world.
I would say:
in a manner that specifies the relevant perceivable properties
The big issue in the debate with Ken was between Andrew and Ken about the extent to which physical properties are perceivable. I agree with Ken that the extent is rather limited, but don't see why that matters if affordance is defined so as to be consistent with whatever that extent is.
And as a Rortian, I would avoid the "mirror of nature" by saying something like:
a properly attuned perceptual system can respond appropriately to perceived properties of the environment
To Andrew and Charles,
ReplyDeleteWell... but... that the cup on my desk is pick-upable is nothing more than that it is a particular size and shape and that I can move and articulate in certain ways. That the lever is press-able seems very similar. That the lever is press-able TO GET FOOD, seems oddly different. Certainly, it is not as natural to use the term 'direct perception' in the latter context. Other terms ecological psychologists are attached to, like 'specificity' also get very messy, for example, if we try to say that the tone is part of the intermodal information that specifies food-get-ability.
As Pam might point out (if she is reading this), you are really in trouble if someone asks why different relationships between the lever pressing and food (in the presence of the tone) lead to different responses. We are in for a world of trouble if we try to assert that a variable interval schedule 'affords' responding at a slow continuous rate while a fixed interval schedule 'affords' a scalloped pattern of response. Such an assertion would bear little resemblance to the concrete observation that the lever affords pressing, or that my cup affords picking-up
Maybe I just have the textbook to heavy on my mind. I certainly think that students need to first get comfortable with the term affordance in what I am calling its most natural usage.
Well... but... that the cup on my desk is pick-upable is nothing more than that it is a particular size and shape and that I can move and articulate in certain ways. That the lever is press-able seems very similar. That the lever is press-able TO GET FOOD, seems oddly different
ReplyDeleteI'm not entirely convinced that this is the case. There are two things you have to do in perceptual learning: learning to differentiate the information specifying the affordance, and learning what that information means, e.g. the pick-up-ability of the cup. You never get to 'peer behind the curtain' of information, so at the level of information, there isn't much difference between 'information that means pressability' and 'information that means food'; if and only if there is a sufficiently reliable connection between the meaning and the information. Typically that connection requires a specification relation to meet this criteria; a connection based in ecological law. In a lab setting, you can make it as if there is a specification relation: a connection based on the experimenter shaping the ecology.
Sabrina's been pushing me on this for a while, since I went over Tony's chapter 6; I'm coming to the conclusion that I agree about the principle, although I still reserve the right to worry that outside the lab, only specification is good enough to sustain and support this learning.
I agree with Andrew that one issue is meaning acquisition (this relates to Pam's comment about training regimes as well). At some point in the past, the rat had to learn that the perceptual information specifying pressability actually means pressability. The lurking question seems to be, does the trained rat come to learn that the perceptual information specifying pressability (because this is unchanged) means food? If so, then the perceived affordance has changed. But, I don't think this is correct. The rat must press the lever to get food. The affordance specified by the perceptual information still means the same thing for action (pressability). So, while I agree with Andrew that we cannot peer behind the curtain (and have been advocating for the idea he describes above for some time), I don't think that idea maps onto this situation. The lever pressing action now has an additional consequence (food) but the information specifying pressability still means the same thing for action.
ReplyDeleteIs is correct to say that the rat perceives this consequence (food) when it perceives the information about pressability? I don't know. I am tempted to say the link between perceiving pressability and food is similar to the link between, say, moving a book out of the way so that I can reach the coffee cup on my desk. To do this I have to know, at some level, that moving a book (effecting an immediately actualisable affordance) will open up the possibility of actualising another affordance that I am more interested in. What is unclear is the status of this knowledge (perception, something else??).
The rat certainly knows something about the food when it presses the lever. At the very least, the positive relationship between lever press and food leads the rat to spend more time near the lever and look at the lever more frequently than without training. If the rat is in proximity of the lever and perceives the relevant affordance, then (if you agree with the idea of compulsory effecting of affordances)this will already lead to increased pressing behaviour without invoking any more sophisticated knowledge about the consequences of lever pressing. But the literature suggests that the rat knows more than that approaching the lever is a good thing. I do like the idea of thinking about the continuous evolution of available affordances, but it is difficult to explain at what level goals (get the food, move the book) influence this process.
Sabrina,
ReplyDeleteYes, yes! You have come to the problem exactly. The information specifying press-ability still 'means' the same thing - that the lever affords pressing.
Given the differences in the physical situations, it seems more natural suggest that there is information in the environment to specify that 'moving the book will allow me to reach for the coffee cup,' than to say that there is information specifying that 'pressing the lever will get me food', but I can see why you think that is a move in the right direction.
As for what the rat 'knows', I'm uncertain. Given my background in Holt-style behaviorism, I would be very comfortable saying that the rats behavior is 'directed at' getting food, and maybe even that the rat 'wants' food. I think the word 'know' might be more problematic than that. What the heck, let's make that the next post - more soon.
"it is difficult to explain at what level goals (get the food, move the book) influence this process."
ReplyDeleteI assume that Gibson addressed a theory of actions as opposed to mere motions such as falling, slouching, blinking. If so, purpose or intention seems implicit in any movement analyzed in terms of affordances. So, if we are to apply such analyses to the rat's immediate movement, that movement must be assumed to be purposeful, even when the rat is just "scurrying about" (the purpose being merely to get from the current position to any other). Then the tone at a minimum narrows the purpose from scurrying about purposefully but aimlessly to scurrying with a bias toward positions from which the lever is pressable.
A continuing problem is how to deal with situations in which a surface does not offer an affordance that is immediately effectable from the subject's current position. It now seems to me that in such situations, reaching a position from which an affordance possibly - but not necessarily - offered by a distant surface could be immediately effected can be viewed as achieving a goal. In which case the tone creates for the rat the goal of getting within reach of the lever so as to be able to effect the "pressable" affordance offered by the lever. (I agree with Sabrina that it was a mistake to suggest that the affordance changes.)
My guess is that "food" as a concept in the mind of the rat plays no role; he has just been trained to respond to the tone by positioning himself so as to be able to press the lever. Of course, I can't prove that, but I'd have to know how the conclusion that "the rat knows more than that approaching the lever is a good thing" was reached before abandoning that guess. Eg, even if he has been observed to do something akin to salivating in response to the tone, couldn't that just be part of the learned behavior?
This way of looking at action leads to the idea of interim goals. Eg, in the coffee cup scenario, the subject's goal is to move a hand so as to be in position to effect graspable. The book impedes the sequence of required moves necessitating an interim goal of getting in position to effect move-aside-able. Similarly, effecting graspable is really just a step on the way to yet another goal, eg, moving the cup to a position that makes sip-from-able immediately effectable. Etc, etc. An effectively endless succession of interim goals and effected affordances.
Charles,
ReplyDeleteAll good issues. In particular, you ask how:
"the tone at a minimum narrows the purpose from scurrying about purposefully but aimlessly to scurrying with a bias toward positions from which the lever is pressable"
Yes, this is part of the quest for a coherent way to talk about what physiology does, i.e., what Andrew and Sabrina keep going on about (http://psychsciencenotes.blogspot.com/2011/08/what-does-brain-do-pt-2-fast-response.html).
What I like about your way of phrasing the question is that it provides a description of the phenomenon that is nicely neutral with regards to the explanation. How is it that the rat's behavior is re-oriented? There are approaches to psychology that assume such reorientation result mental representations, i.e., the tone makes the rat think about food. When I am in nicer moods (which apparently I am in now), I would say that there is nothing wrong with such claims, as long as the person making those claims remember they are hypotheses. Of course, that is a historically and developmentally contextualized statement on my part. There is nothing wrong with my students doing it, and nothing wrong with past psychologist and philosophers doing it, but at this point professional psychologists should be willing to reject the hypotheses.
Sigh. I feel like I am losing coherence. More later.
Eric -
ReplyDeleteNotwithstanding my neutral phrasing, I'm not really neutral on representations. That's why I resist Sabrina's use of "concept". I have been influenced to think of a concept in terms of the use of a word, and of the words likely to be relevant in this context as referring, which suggests the possibility of a mental representation of the referent. Any or all of which may be a matter of confusion on my part - and at least some of it must be wrong since Sabrina rejects representations.
The information specifying press-ability still 'means' the same thing - that the lever affords pressing.
ReplyDeleteI'm as yet convinced. The 'same' information variable (ie the same geometric pattern in the optics) is going into a different measurement device in the two cases; into the 'rat pressing lever' device, and the 'rat pressing lever for food' device, which is what the former turns into after learning. The meaning of the information can, I think, change this way.
This is where it gets utterly crucial to keep the world, the information about the world, and what that information specifies rigorously separated (at least analytically).