In an earlier comment, I indicated that I would post something on my blog to help clarify the distinction between direct and representationalist perception.
I have now done that, in a series of four posts. The first of those is
and it, in turn, contains links to the other posts.
Your turn.
I like to refer to “natural selection” as “environmental design” to claim back the word “design” from its subversion into some form of magic mantra. It’s a personal foible. Take no notice.
Well, I was hoping for at least a five minute argument if I can’t get the full half-hour. 😉
http://en.wikipedia.org/wiki/Benham%27s_top
I sympathize — er, sympathise — but by using ‘design’, you run the risk of implying that something teleological is going on.
Let me decide what to charge. I need to adjust for inflation and do a currency conversion.
Neil,
Here’s a GIF of the illusion. I created it manually, using the Windows Snipping Tool, so the images aren’t perfectly aligned. In the original, the green square is stationary, while in my version it appears to move slightly. However, you can still see the perceptual effects in my imperfect version.
http://i59.tinypic.com/302yjd0.jpg
Neil,
I agree, though I emphasize that the judgment isn’t a conscious one in the case of perception.
Check out the Ternus illusion. This is a motion illusion in which there are two competing possibilities, described in the article as “element motion” and “group motion”. Which one wins depends on the duration of the blank “interstimulus interval”, aka “ISI”, between the two static images.
With shorter ISIs, the visual system judges that element motion is the better inference. With longer ISIs, group motion wins.
Not only is the motion itself an inference, but even the type of motion is an inference.
I disagree. Neurons are slow, but they’re not that slow, and besides, neural networks are massively parallel.
I hope walto does comment further, preferably in sentences that use more terms of more than one letter. Although, as an ex-mathematician, I am embarrassed by asking for less formality.
I don’t know if he will agree with your paraphrase, but I will try to address it directly, mainly for selfish reasons: to see how well I understand the philosophical issues by trying to write something coherent.
First, the word “non-physical” in your paraphrase: I believe the philosophy is talking about properties of the mind, not the brain. The argument is meant to apply regardless of one’s position on reducibility of the mind to the brain, so I think we need to avoid any appeal to saying something like “if it is a property of the mind, then it must correspond to a property of the brain.” So non-physical means in the mind, I think. (Note that this does not prevent one from appealing to the mind supervening at least partly on the brain in arguments that appeal to physicalism, ie token identity, not type identity).
Second, “apprehend”. I take to mean awareness from a first-person perspective. So the explanation of perception must speak to the phenomenology which I will take to mean our old friend, qualia.
A third aspect to consider is why the explanation of perception is considered to be direct or indirect. Please note that I am trying to use these terms as philosophers use them, not as psychologists use them.
By perception, philosophers mean
Three schools of thought:
Sense data: This is indirect perception. We perceive objects called sense data which have the properties of the physical objects they correspond to, eg greenness. The sense data are what we perceive, not the objects themselves, hence indirectness. Illusions and vericidal perception share the same sense data but these do not correspond to the same physical object. A criticism of sense data is that they do not seem to supervene on the physical.
Representationalism: This is direct perception. Perceptions are representations, ie intentional. We don’t perceive representations, the perception is the representation, so this is considered direct perception. Qualia are just a brute fact of this type of representation. Representations supervene on the brain. Illusions are misrepresentations.
Disjuntivism: This is direct perception. We perceive the object directly. The perception supervenes on the physical, but the physical includes both the brain and the object being perceived. Qualia result from the dual supervenience. Illusions share the physical tokens (not properties) of the brains states of the vericidal perceptions, but are not the same type. (This is the source of the name disjunctivism: perceptions / illusions are this type OR that type, but do not share some type, as they do for sense data or representations). (I’m being a bit careless with “type” versus “property” but hopefully my point is reasonably clear.)
The sense data account is mainly discounted by philosophers today due to its problems with physicalism. Representationalism is popular but explaining qualia is a weakness. Disjunctivism is less intuitive (at least to me) but some take it as giving us the advantages of sense data without their problems of dualism. Plus it may be thought more consistent with embodied perception for fans of this.
All of these explanations address illusions from a philosophers viewpoint. At least for initial understanding of the position, the psychological and neuroscientific levels of explanation are not relevant.
Thanks. However, I still see no basis for your claim that there is inference. It still seems entirely compatible with direct perception.
It’s perhaps reasonable to say that construction of transducers involved statistical inference. But that’s quite different from your claim that actual acts of perception use inference.
Generally, those who have denied that perception of things like tables and chairs is direct have claimed either that the existence of those things is inferred from non-physical stuff (like sense-data) that IS directly perceived or that those “things” are nothing but “logical constructions” of (again non-physical) sense-data. If what is directly perceived IS in fact physical–just neither table nor chair–then there is direct perception of physical objects which is what the direct realist was insisting on and her critic was denying in the first place.
You didn’t get the biz about in or around quite right. It’s not that what actually is apprehended can’t be in or around the perceiver, it’s that having some causally necessary process for the perception of some physical object in or around the perceiver isn’t sufficient to make direct realism false. You need to add the requirement that something is apprehended.
Hope that helps.
BruceS,
Keep the context(s) in mind.
When direct perception first came up, KN and I were discussing its implications for idealism:
In that context, even causal indirectness is enough to maintain Berkeley’s idealism as a live possibility, which is why I brought up the “photons coming from a hologram of green leaves” example. If perception is causally indirect, then every stage of the causal chain is subject to virtualization by Berkeley’s God. Perhaps the leaves aren’t real, but the photons are. Perhaps the leaves and the photons aren’t real, but the retina is. Perhaps the leaves, photons, retina, optic nerve, and brain aren’t real, but the percepts are. As long as the observer’s mind is real, and the percepts are presented to it, then idealism remains possible.
That’s the first context. The second context is this thread itself, in which Neil is defending his version of direct perception. Neil insists that perception does not involve inference or representation, so I have focused on visual illusions that demonstrate the opposite.
We can expand the discussion to include other questions about perception, of course, but so far I have been trying to focus my comments on the thread topics themselves.
Neil,
If so, then there should be a description of the process that does not invoke inference or representation. Can you supply one for the Ternus illusion, in both the “element motion” and “group motion” cases?
In the Ternus illusion, the visual system is weighing two hypotheses against each other and judging one to be superior, based on the interstimulus interval. At low ISIs, element motion is perceived. At high ISIs, group motion is perceived. That fits your description of an inference perfectly:
That’s not quite right. That perception is direct is an hypothesis. I am not asserting that it is settled fact.
Neil,
We’re evaluating this claim of yours:
walto,
Direct perception of an intermediate is indirect perception of the intentional object. That’s what ‘indirect’ means.
We perceive the intermediate, which is why we can see motion where there is none, and why we “know” that the red ball is behind the green square when there is absolutely no way of directly perceiving that fact. It must be inferred.
That’s not what you said earlier. You wrote:
When you emphatically state that the intermediary does not TAKE PLACE in or around S, I have to assume that you mean what you say.
If you allow that the apprehension can be subconscious or preconscious, then the causal intermediate qualifies, and direct perception is falsified. If you insist that the apprehension must be conscious, then it seems to me that you are begging the question. There always has to be some first conscious apprehension in the causal chain, and a “conscious apprehension” will always qualify as a perception.
I used the expression “more likely”. That leaves it as hypothesis, not as a truth claim.
You keep on coming up with examples that show zero evidence of inference. And you keep on insisting that they demonstrate inference.
There is likely something that you are seriously misunderstanding about “direct perception” but I’m not sure where that misunderstanding is.
Neil,
Right, and we are evaluating your claim. Is direct perception “a more likely candidate for explaining human perception”, or isn’t it?
I provided an example that fits your own description of what an inference is:
Do you disagree with your description? If so, what is your new description, and what is the justification for the change?
On the other hand, if you stand behind your description, then how does my example fail to qualify? Please be specific.
If I am in fact misunderstanding direct perception, then it would be very helpful if you would answer the question I posed:
No, you didn’t.
Rather, you provided an example that you claim fits my description. But there’s no evidence, that I can see.
Neil,
As I said:
All of your examples are about the subjective experience of perception. The direct perception hypothesis is not about subjective experience.
Neil,
Sure it is. Whether direct or not, perception is inherently subjective (though the details of how it operates are not). Two people can look at the same thing and perceive it differently. You might look at the second illusion on this page and perceive the balls moving vertically, while I see them as moving horizontally. It’s subjective.
A good way to proceed would be to compare competing explanations of the Ternus illusion. I have provided an explanation in terms of indirect perception. You claim that direct perception is a better explanation. I’ll ask again:
OK, sorry for going off topic.
I personally find KN’s contibutions more interesting than any link to considerations of Berkeley. I was trying to engage with them on their own terms; it is true I ignore the Berkeley stuff.
I’ve already commented to Neil on how I think the three levels of explanation present in his posts need to be addressed separately; my post was more about the philosophical level.
But I’ll stop now.
I do have one question, though: do you happen to know if Dr Liddle gives out prizes for longest posts? If so, where can I submit my entry?
BruceS,
No need to apologize. I wasn’t scolding you, just explaining why I was keeping my comments focused on the mechanics of perception.
Feel free to bring up the topics here, or in a new thread if you feel like posting one.
If that’s an oblique way of asking about posting privileges, Neil or Alan can grant them to you. I look forward to seeing an OP from you.
Actually, it was an oblique way of making a bad joke.
Thanks for the posting suggestion — I don’t have anything original to say, however.
keiths writes: “Direct perception of an intermediate is indirect perception of the intentional object. That’s what ‘indirect’ means.”
Exactly. I’d just add that if one believes that this intermediate object is physical, one would pretty much have to admit we have direct perception of physical objects.
Also, I did indeed write:
“What epistemically relevant intermediary do you think must be apprehended by S (not TAKE PLACE in or around S) in order for the the motion illusion to work for S?”
And I see it was not sufficiently clear. I should have written “What epistemically relevant intermediary do you think must be apprehended by S (not just TAKE PLACE in or around S) in order for the the motion illusion to work for S?”
I am happy to accept that revision, as I think should have been clear from my last post on this.
Yeah, just casually following the dispute between KeithS and Neil Rickert, was wondering if it would make sense to put the discussion in the context of an automated vision system (of which countless examples exist in our modern age), wherein you would have some sophisticated sensor(s), with all sorts of capability built into hardware logic of the sensor itself, and then on the backend, more processing capability, of a programmatic nature, residing in RAM, etc. I guess one would characterize the backend as “inferential” and the front end sensors as “direct”, but one could always identify repetitive back end tasks and transfer their functionality to the hardware sensors themselves, at which point those tasks become “direct” (as opposed to inferential). And it seems that even the hardware sensors could be fooled by optical illusions without characterizing their behavior as inferential. Or, possibly validly characterized as inferential, as isn’t this dispute possibly semantic in the final analysis, don’t know.
[edit] A few other things to throw into the mix — what about directly touching something and then inferring what it is — is that relevant to anything. Maybe the first time you touch it you infer something through reasoning, but then if forced (in some scenario which I can’t really articulate at the moment) to identify by touch the same item 1000’s of times you start directly identifying the object without any deliberative process at all. But all that aside, everything is mediated through sensory data given off by an object, yes? I mean we’re not entering into some shared direct experience of the object’s reality merely by perceiving it, right? I’m inclined to say everything is inferential — we don’t confirm omnisciently anything by merely perceiving “something”. (What am I talking about, I don’t know).
I mentioned a couple of examples in the posts at my blog. Specifically, I mentioned the barcode scanner at stores and the ethernet card. In both cases, there is inference following the basic direct processing.
I agree that, in real life, there is inference. And I already mentioned that I find it hard to say where perception ends and inference begins. I guess I would say that I see perception as relatively thin, and we add a lot — perhaps at an intuitive level that might not count as conscious thought but could be considered inferential.
walto,
It seems strange to call it an “object” when we are talking about a pattern of neural firings. It’s physical, but it’s not really an object.
Two points in response:
1. I haven’t claimed that all perception is indirect. I’ve merely claimed that it can’t all be direct, and I’ve given examples that support my claim. Direct perception cannot explain the motion illusions I’ve cited.
2. We need to be careful about language. In indirect perception, we aren’t really perceiving the intermediary; what we’re doing is perceiving the object through the intermediary. The distinction between direct and indirect perception rests in whether intermediaries (in the form of inferences, representations, or models) are necessary. Neil says that they aren’t, and that perception is therefore direct. I say that you cannot explain the motion illusions without an intermediary. After all, you can’t directly perceive motion that isn’t even there. An inference is required.
I addressed that already:
JT,
That’s why it’s better not to define “inferential” as being associated with the back end. Designs can be partitioned in all sorts of ways, and there’s no essential correspondence between the back end and inferential processing.
Inference is a function, and anything carrying out that function is inferring. Neil provided a description:
That is exactly what happens with the Ternus illusion. As I remarked earlier:
Neil,
If you concede that subconscious or preconscious inferences happen, and that the result is a conscious experience of motion, then you have conceded that perception is indirect in the case of the green square illusion.
The only way to avoid that conclusion would be to redefine perception to exclude inference, representation, and modeling. But then “perception is direct” would become a mere tautology containing no information about the real world, and you would be using the word “perception” in an idiosyncratic way that clashes with its normal meaning.
This “perceiving through the intermediary” does not constitute indirect perception on my definition (or others I’m familiar with), though no doubt there are a lot that I don’t know.
Spinoza said “the mind is the idea of the body.” But that could mean several different things, so it makes sense to disambiguate that “idea of”. And that’s what direct realists generally try to do. So, unless you’re saying that there must be an intermediate [i]object of perception[/i], you’re not saying anything that contradicts most modern formulations of direct realism.
I understand that you say there may be subconscious apprehensions of intermediate objects. Direct realism will be shown to be false only if there’s a good argument that there MUST be. I don’t think there are any such arguments.
This thread makes me want to start a new thread titled Is Philosophy Obsolete?
No, I haven’t conceded that. When we concentrate on what we are looking at, we see more detail than without the concentration. But I would count that as conscious activity.
Thus far, your “argument” amounts to a bare assertion, much repeated.
walto,
What do you think “indirect” means?
If I say I indirectly caused something, I mean that I caused it, but through an intermediary. If I say that I indirectly perceived something, I mean that I perceived it, but through an intermediary.
Neil has specified that he is arguing against intermediaries in the form of inferences, representations, and models. My counterargument has been to present a number of illusions that require such intermediaries.
My challenge to Neil extends to you or anyone else who wishes to defend direct perception:
Of course perception is causally indirect. (Almost) nobody denies that. (The “almost” is there for guys like Travis and Hyman, who deny perception is causal in the first place.) If all you mean is that there are causal intermediaries to perception, you will not be saying anything very controversial. In any case I will certainly agree with you.
So the question is whether you want to claim that perception is indirect is some manner that anybody doubts.
Neil,
You don’t have to concentrate on the motion illusions to see the illusory motion.
Hardly. I’ve presented specific arguments for why the illusions I’ve cited require inferences, and therefore must be instances of indirect perception. Do you have a counterargument? How do you explain these illusions in terms of direct perception?
Let’s compare the theories side-by-side and see which one better explains the evidence. That’s how science works.
walto,
Did you read my entire comment? It included this paragraph:
Yes, I read that. I argued in the paper I linked to above that even seeing what mirror images reflect requires the apprehension of intermediaries (the images). That was my main complaint about disjunctivism. So if you think that perception of physical objects is not always direct, I agree with you.
walto,
But then you proceeded to ignore it in your next two sentences:
Mirror images are merely causal intermediaries, and causal intermediaries are not under dispute.
The topic being debated is whether intermediaries are required in the form of inferences, representations, or models.
Again, my challenge to Neil extends to you or anyone else who wishes to defend direct perception:
I argue that mirror images must have more than causal status in my paper, so your first remark is wrong (as well as weirdly nasty). Not sure why you keep taking this “Did you read that?” tone, but it’s not very attractive.
And I think what you and Neil are arguing about may be more fun for you two than for me, thanks. For reasons I’ve already repeated a couple of times, I don’t think your “challenge” is relevant to any question of direct perception in which I have much interest.
walto,
Pointing out that you ignored my statement is “weirdly nasty”?
This isn’t something that Neil and I are just making up. It’s the standard debate between proponents of direct and indirect theories of perception, as Michaels and Corello explain:
You engaged that topic when you wrote this:
I’ve argued that the motion illusions do require a preconscious inference — an intermediate. If you, Neil, or anyone else can offer a counterargument, I’m happy to consider it.
No, you haven’t. Your argument comes across as a strong case of confirmation bias.
And maybe there’s some argument from ignorance in the mix – “I can’t think of how this could be done without inference, so there must be inference.”
I would certainly welcome it if you ever felt inclined to post something about the how the questions that philosophy (as opposed to science) of perception is trying to answer and why you see them as interesting/important. Also, based on the your statement of views on your Ithaca pages, why you are a naive realist as opposed to the other positions.
I did read the paper you linked, but I have to admit the focus was a bit narrow and the required philosophical background a bit too much for me to understand it deeply.
It is true that this topic of philosophy of perception may not be of interest to many TSZ readers.
But I also noticed from the Ithaca page that you leaned toward moral realism. Based on some of the other discussions here, I suspect a post about why you hold the position of moral realism would attract a lot of interest.
Keiths writes: “This isn’t something that Neil and I are just making up. It’s the standard debate between proponents of direct and indirect theories of perception, as Michaels and Corello explain…”
I haven’t read that book, and it may be of great interest/importance to cognitive psychologists. I did glance through the index, though, and didn’t notice any philosophers in there. That’s likely because the focus is different. I don’t know enough about that field to say much more. But the manner in which motion illusions are perceived or understood is not anything I know or care much about.
You have again misunderstood my point (maybe I wasn’t clear enough again?) when I said a refutation of direct realism would require that there must be intermediaries. That wasn’t intended to mean a showing that there are some sorts of perception that require intermediaries. (I’ve already said I agree with that position.) It was intended to convey the necessity of a showing that NO physical objects can be directly perceived. (And again, I generally don’t mean by “direct” what you mean by “direct.”) Direct realists say that physical objects can be perceived directly. Others deny that. The fact that some motion illusion requires the perception of an intermediary is neither here nor there.
And yeah, I think you’ve been weirdly hostile in several of your comments. I don’t know you or this place, though. So maybe that’s the general tenor here. I hope not.
Bruce, your question of why I think the philosophical questions are interesting and/or important is a hard one. It may be just an accident of my history. You know, if you take a bunch of philosophy classes and/or read a bunch of philosophy books you’re likely to have (get?) an interest in different issues or disputes than if you’ve studied optics or physics or cognitive psychology. It’s probably kind of a chicken/egg thing.
I certainly understand if most of the members here (and others) find the philosophical issues arcane and pointless. That’s kind of an occupational hazard of an interest in philosophy.
When you refer to my Ithaca College connection, are you talking about academia.edu? I was an undergrad at IC a hundred years ago, and taught there for a couple of years after I got my Ph.D. I’ve lived in the Boston area for a long time and hope to be doing some evening adjunct work at a local university here this fall. If so, I’ll change my affiliation on that site.
I don’t remember saying anything about moral realism anywhere, but I tend to be an Everett Hall groupie and he was a realist in that area as well as elsewhere, so I probably was just indicating my me-too-ism. I haven’t thought much about ethics for a long time, but the course I think I’ll be teaching is Intro to Ethics, so I’ve been feverishly reading or re-reading a bunch of stuff in that area. If I think of anything interesting (and I have time) I’ll try to post something. But I have to say the atmosphere here seems to me a smidge angry.
Neil,
keiths wrote, “I’ve presented specific arguments for why the illusions I’ve cited require inferences, and therefore must be instances of indirect perception.”
And you responded, “No, you haven’t….” Do you take the position that there are no indirect perceptions? Thomas Reid held something like that at one time, I think. But he also said things that seemed to conflict with it. Anyhow, I think it’s tough road to insist that NO perception requires either inferences or epistemic intermediaries, and I can’t tell if you’re taking that (to me, kind of radical) position.
Edit: I guess another way of putting my question to you is Do you think (a) Keith hasn’t presented any good arguments for the view that this type of perception is inferential, or (b) even if he did, it doesn’t follow from the existence of that sort of “inferential perception” that perception is ever indirect? Or…..what? Thanks.
Neil,
My argument is that direct perception is incapable of explaining perceptual phenomena such as the motion illusions. If I’m a victim of confirmation bias, then you’ll be able to demonstrate that by giving a clear explanation of how direct perception does explain the motion illusions.
If we are comparing two theories, one of which fits the evidence and and the other of which doesn’t, why should we prefer the one that doesn’t?
On the other hand, if you think that direct perception does explain the motion illusions, then tell us why and how. If your argument is persuasive, then we’ll all be able to see that direct perception is the better theory, and that my own position is just an expression of my confirmation bias.
BruceS,
I would welcome it too. We spend a lot of time discussing ID and theism, but Lizzie certainly didn’t intend for discussion to be restricted to those topics. We have a “Philosophy” category, and it has quite a few posts in it. Some posts on the philosophy of perception would be a good addition.
Folks who aren’t interested are free to ignore them, of course.
No. At present, all I can say is that direct perception seems to fit reasonably well. But people don’t all mean the same thing by “perception”. And if somebody is using a remote TV monitor to observe, I suppose that would be best described as indirect.
I’ll go with (a). Keiths repeatedly gives examples that are fully consistent with direct perception, yet claims that they require inference.
That’s your assertion. The “argument” part is missing.
What could be a more direct source of information about motion, than an apparent change of position between to time instants? That’s close to the first-principles definition of motion. Yet you assert it requires inference.
According to Gibson, we get information by means of transducers that are tuned to particular invariant features. It follows that it should be possible to fool the transducers by providing just those features, but omitting everything else. So direct perception predicts that there can be optical illusions. Yet you assert that illusions pose a problem.
Thanks Walt. I was referring to this page. Hopefully, I have the right person!
My philosophical views
I may be way off base here, but I think sometimes the angry tone, which is how some posts come off for me too, comes from the background of some of the posters here having had to deal with posters at creationists or ID sites. (But not in general those people who believe in ID and who post here).
In any event, there are many good reasons to post. One certainly is because you enjoy the cut and thrust of intellectual duels.
But that is not my reason.
I read and sporadically post simply to understand and try to engage with the thoughts of the many interesting people who post here, especially those whose background is foreign to an IT person like me. I personally find philosophy interesting for its own sake, and try to deal with it as best I can on its own terms. So that is why I am sure I’d find anything you care to post interesting.