Given that the school year at CMU starts today, this will be my last post (split in two). Thanks to readers for very helpful discussion in the previous posts. Part 2 will be published late Tuesday or Wednesday. [Update: teaching classes today, so will be slower to respond, though I did finish writing my lectures!].
I’m not going to discuss the visual streams as planned. Still, some may be interested, so to voice the claim I would have argued for: parts of the dorsal stream play a central role in visual consciousness. The dorsal stream is not a purely unconscious stream (see here). Ok, moving on to today’s question:
What is introspective attention?
Maybe not what many philosophers have thought.
What follows is work in progress, so there will be many gaps. But I hope there are some new points to be made in what follows on introspection and transparency. I know there are readers who know much more about these issues then I do, so I will be grateful for help!
Attention has a central role to play in philosophical investigation of introspection and of phenomenal concepts, yet philosophers often have little to say about attention even as their theories invoke the idea at crucial points. There are exceptions, of course, among them, Brie Gertler whose work here and elsewhere I strongly recommend.
Let’s focus on introspection of perceptual consciousness. It is very natural to think that introspective attention is simple in its nature, for there isn’t much more to it (at its core) then that it gives us direct first-person access to the phenomenal features of consciousness. In doing so, it can allow for a sort of (near) infallible thought about consciousness. These points are compelling, captured in talk of acquaintance.
But I am suspicious of these ideas.
My goal is to contrast two conceptions of introspective attention: the direct model that gives rise to talk of acquaintance and the transparency model that is altogether different. On the nature of phenomenal properties, choose whichever theory you want, but I it will be helpful to think of them as non-physical properties to maximize the contrast between the two models. OK, here we go.
Transparency
Gilbert Harman notes the following:
‘‘Look at a tree and try to turn your attention to intrinsic features of your visual experience. I predict you will find that the only features there to turn your attention to will be features of the presented tree’’ (1990, 39).
Call this the transparency observation. The significance of it has been debated: Does it legislate against qualia, intrinsic features of experience? Does it establish or give evidence for representationalism? I have my doubts, but these are not our topics. Here’s what I take from it:
The capacities called upon in perceptual attention are the same as those called upon in introspective attention.
To see why this emphasis is different from what is generally discussed in the literature, consider two distinct readings of the transparency observation:
- Object-Centered: The target of introspective attention is the same target as that of perceptual attention.
- Process-Centered: The process of introspective attention and perceptual attention are the same.
I take Harman to focus on the first (what you attend to in introspection is what you attended to in perception), though transparency adherents endorse both. Given their targets (establishing representationalism, eliminating “qualiaphilic” theories), transparency adherents have emphasized the object-centered version, though the two versions are logically independent. On the other hand, opponents of transparency deny both versions, though in engaging their transparency opponents, they focus on the object-centered version as well.
I think this emphasis has obscured some crucial points, and I think there is new insight to be gained by focusing on the process-centered version of transparency. If we refocus transparency then the question becomes:
Is there a single process shared between perceptual and introspective attention such that there are interesting philosophical implications?
There’s Attention and then There’s Attention
There are two stages in introspection where we can correctly talk about attention: (a) the process that fixes introspective thought or (b) introspective thought itself. In general, I shall speak of (b) as cognitive attention and (a) as attention that informs such thought (it could be perceptual or introspective attention). Talk of cognitive attention might sound odd to some of you, as the focus in attention theory has been on (a), but even William James noted in his famous description of attention that thought is a form of attention. We can think of it as a conceptual form of the directedness and selectivity that we often ascribe to attention.
On (a), we have two putative types of attention: perceptual attention informing empirical thought and introspective attention informing introspective thought. So, our question again is this: are they the same process or different?
Direct Theories
Direct theories (or acquaintance theories) are so called because they claim that we are directly aware of phenomenal properties by attending to them. But can these theorists say more about the nature of this form of attention? Where philosophers have provided an answer, it seems clear that introspective attention is different from perceptual attention as investigated by cognitive science.
Here are some properties that I have taken from Brie Gertler’s discussion (see also Dave Chalmers’ discussion for similar claims):
- Introspective attention is direct.
- It thus allows for a special way of thinking (conceptualizing) phenomenal features, perhaps a special type of demonstrative thought.
- The attended phenomenal feature is embedded in the resulting thought.
Gertler’s view is nuanced, and these bullet points don’t do it full justice, but they capture the basic idea. It is the embedding that helps to explain the distinctiveness of concepts of the phenomenal and some of their epistemic properties. Since perceptual attention does not embed, and since this is a crucial feature of introspective attention, as Gertler and Chalmers I think would emphasize, then we do have a distinct form of attention. I suspect that attention as characterized here is what many philosopher have in mind when they speak of introspective attention. If not, then I ask them: what do you have in mind? Please do say…
The Challenge
Why believe that there is any such thing as direct attention that allows for embedding? Two possibilities: (a) theoretical constraints from elsewhere require that we postulate such a faculty and (b) something like introspection itself reveals it. Because philosophers often seem to suggest that we are obviously in a position to directly be aware of phenomenal properties, I want to first question (b), and return to (a) in the second part. For I assume that they are relying on introspection itself to make those claims.
First, a brief aside based on personal experience: As an exercise, find a scientist who works on perception and who is not already up on the vocabulary that we use to describe consciousness. Better if they don’t already work on consciousness. Now, try to get them to recognize that they can directly attend to phenomenal character. I predict that this will be quite hard to do which, if my prediction is correct, reminds us that attention to phenomenal properties is not as obvious as we might think it is. End aside.
Start with Harman’s observation: attend to some physical feature, F, of an object and think about it. Here, you are having a perception-based empirical thought about the world mediated by attention selecting F. Now, turn attention inward: introspect the visual phenomenal character Φ associated with experiencing F, and on that basis, have an introspective thought about Φ. Is there really attention that embeds Φ?
We noted that there are two stages where talk of attention gets a grip: thought and the selective process that leads to thought. When we transition from perceptual attention that serves empirical thought to introspective attention that serves introspective thought, we can note three uncontroversial instances of attention: the perceptual selectivity at the first stage, the empirical thought that it gives rise to, and the introspective thought in the switch.
OK: To the exercise again, and toggle attention back and form between attending to F and then to Φ. Is there really anything that you are are clearly toggling except cognitive attention, namely moving between introspective and empirical thought on the basis of a selective capacity? This is the process-based transparency point.
The transparency theorist has the following suspicion: either direct model theorists identify something that is not there in self-reflection (attention that can embed phenomenal features in judgments) or they are in fact speaking of cognitive introspective attention. Both options are consistent with transparency. So, the question then is whether we have any other reason to postulate the direct model…
Hi Wayne, thanks for another very interesting post! I have really been enjoying your contributions to the blog as well as the ensuing discussions! Thanks also to John for initiating this great new feature!
I really like the idea of cognitive attention. I think I have been waiting to find out that others have had this idea since I have come to something like it in thinking about the relation between higher-order thought theories of consciousness and attention theories like AIR. For instance Rosenthal argues that introspection involves a third-order thought about the second order thought and also that introspection involves being conscious of your mental life in an attentive and focused way. The only thing could mean is that the third-order thought amounts to something like cognitive attention, but then ordinary higher-order thoughts should (on the HOT theory) count as a kind of cognitive attention as well…I take it that this is in line with your claim that the two processes are the same but interpreted in terms of higher-order theory. It may be a way to unify this group of theories. I know this isn’t really on the topic of your post but I would be curious to here what you thought about this.
Hi Richard:
Thanks for your nice comments. I was hoping you’d weigh in at some point, so here we are!
Interesting: can you remind me where David talks about attention in introspection? I’d like to revisit it.
I think that the HOTs will, in virtue of being directed thoughts, count as forms of cognitive attention, at which ever n-order they are. But I wonder if HO theorists have something else in mind when they speak of attention, though one of my questions is: what do they then have in mind?
That said, I have to ask how one forms a n-level higher order thought about a lower order thought at the n-1 level. You might think that the n-level thought must be attention based itself on the n-1 level thought, in which case, there must be a second notion of attention that is in play, a form of attention that “couples” the n-1 thought to the n level thought.
This “coupling” effected by attention is actually the subject of part 2 for Wednesday, but I’d be happy to email you my draft of that post separately or you can wait for the finale!
But to raise the issue clearly for this thread (to HO theory folks): do they think that there’s an additional form of introspective attention beyond cognitive attention? If so, what is it?
A crucial point in this post is that we have to start talking about attention if it plays a central role in our characterizations of philosophical phenomena like introspection, HO theories, phenomenal concepts etc.
Thanks for the quick reply!
David doesn’t really ever talk about attention anywhere in print but he does say, in his ‘Introspection and Self-Interpretation’ essay which is Ch. 4 of Consciousness and mind, that when one introspects one is conscious of one’s own mental states in a way that “seem attentive, deliberate, focused, and reflective” (page 110 CM). But as far as I know about his views he thinks that attention is completely independent from consciousness. I think he would claim that there are many reasons that we come to have higher-order thoughts, attention is sometimes the answer but it is sometimes other things. So I don’t think the higher-order view needs to invoke a second kind of attention, but they could.
Speaking for myself, I tend to see the connection generally going the other way in that I tend to think of consciousness as usually directing attention. I haven’t really thought it through all the way (though David and I did discuss it at his session in the first online consciousness conference back in 2009)…so, yeah, I would be very interested to see the material for the next post. If you have the time to send it my way before Wednesday that would be great, otherwise I’ll just wait to see it then.
Hi again Wayne,
I’d like to second Richard’s thanks for all of your recents posts–I’ve been following with interest even when not participating.
I’m sympathetic to your claim about embedding in introspective attention, but wonder about the analogous claim for perceptual attention. Perhaps I haven’t quite understood what embedding is, but it seems to me that acquaintance based singular thoughts can embed features of the environment selected by perceptual attention, e.g., I’m thinking about *that* shade of blue (the trope, not the universal). In such a case, I can imagine arguing that there’s a subtle shift in the phenomenology when toggling between perception and introspection: the property is experienced as particular vs. abstract. Now, you might not like tropes, or acquaintance based singular thoughts, but since some people accept them, I was wondering if you could say more about the claim that perceptual attention does not embed.
Thanks!
Thanks James. I am not sure exactly what embedding comes to, but I think that your example might be a way of thinking about it in introspection.
Perhaps one has a view about singular thoughts where this involves object-dependence: no object, then no thought (and not just a false thought). One might then claim that the explanation of object-dependence is that the singular thought in some way has the object (or trope) as a constituent. If one can make sense of this, then I can imagine the direct model to simply endorse this for the introspection case: attention in introspection does the same thing for singular thought about phenomenal properties.
Actually, I think you were inclined to drive a sort of wedge between the two cases?
On whether perceptual attention embeds. I am inclined to think that perception-based demonstrative thought depends on attention, and that there is for such thoughts an object-dependence: no object, no thought (here, I gesture helplessly to Evans via Varieties of Reference, and mutter, “What he said…”).
But isn’t it an additional step totalk about embedding, where that means something like the object becomes a *constituent* of the thought and that attention explains how it becomes a constituent. If that’s the picture, I don’t have an argument against it, but I do wonder what is gained by those additional claims.
I take your point: I assumed that perceptual attention does not embed, but there are some pictures where it does.
Good point that object dependence doesn’t entail embedding (I find this term helpful, if I understand it correctly to mean that a state that is partly constituted by an object or feature making that object or feature available as a constituent of a further state downstream). Since I’m already sympathetic with object transparency, I think the idea of process transparency (cognitive attention toggling between thoughts about what is perceptually available and thoughts about one’s experience) is also helpful for thinking about what’s going on in this context.
A follow up question: do you think there is phenomenology that is distinctive of the toggling e.g., what it’s like to form thoughts based on introspection as opposed to perception? If so, do you have a suggestion for accounting for this independently of appeal to Φ?
Hi James:
Interesting question. I’m still not completely sure about how to argue for cognitive phenomenology (there are attempts to do so, as you know). I have appealed to a broad notion of cognitive phenomenology to explain the phenomenology of attention in my “What is attention?”, where the phenomenology of demonstrative thought tied to our awareness of what we are attending to explains the distinctive “focused, spotlight” phenomenology of attention. So, I’m not against the possibility of cognitive phenomenology.
One obvious response then is to appeal to the thought to explain the difference, what it is like to think about the world versus what it is like to think about one’s experience, the toggling phenomenology being the toggling of thought. This raises the question of the exact content of the introspective thought which involves, presumably, a phenomenal concept. I’m inclined to say that the phenomenal concept is a descriptive concept though it has a demonstrative element which, in veridical experience, refers to the world.
But that’s as much as I want to say in public, since it really is just a suspicion I’m driven to, once we drop the direct model.
It is open to the direct model account to acknowledge toggling cognition as the source of the toggling phenomenology, if one thinks there is such a phenomenology (and perhaps there is). In this case, they would appeal to PHI.
One thing I didn’t comment on in your first post, was your suggestion of a possible difference in phenomenology between attending to the trope and attending to the universal. I’m not sure what to make of the possibility and you were raising it, not necessarily endorsing it. What I am inclined to say is that at some point, the appeals to phenomenology are not going to get us far (this was (b) versus (a) in my post). You probably will agree?
Hi Wayne,
Thanks for this–very interesting. I’m still trying to understand what cognitive phenomenology is, but I think this kind of appeal to it is helpful and plausible.
I definitely agree that phenomenology by itself provides little evidence, though if a view “gets the phenomenology right” that might speak somewhat in its favor. In this instance, however, I don’t think the phenomenology is easily characterized, which might undermine the force of appealing to it at all, or perhaps invite us to describe it more clearly (as your toggling example does).
I should add that I am not against the appeals to phenomenology as providing information and evidence. But I think that we are just going to find a bit of disagreement here such that we will be pulled in two if not many directions (i.e. the phenomenology of introspection). My appeal to phenomenology about suggests that it is equivocal, across subjects, on the phenomenology of introspective attention. I share the transparency intuition in a sense that I spelled out a bit in the post; others don’t share that intuition at all. SO, we have to find another way to get some traction, and hence, Part 2, tomorrow…(cue suspenseful music).
Hi Wayne. Another fascinating post! Jakob Hohwy has a recent paper out “Phenomenal Variability” where he discusses the relation between introspection and the default network, how asking subjects Harman’s question, in effect, forces them to apply introspective attention in an ‘unnatural’ way (perhaps explaining the difficulty nonphilosophers face when asked ‘please introspect’ in certain contexts).
But for, me the elephant in the room in all these considerations involves the *heuristic* nature of the capacities involved. Given the sheer complexity of what’s actually going on in the brain, it seems like the prior question must be, How much information of what kind should we expect these introspective attentional processes to have available? We have good reason to think metacognition is drastically heuristic, which is to say, to involve domain-specific problem-solving devices. The question then becomes one of applicability. Does it even make sense to expect answers to the kinds of questions you are asking?
Hi Scott:
Thanks for this thought. I’d be interested in understanding Jacob’s point a bit better. I guess a natural response, to those who are friendly to the idea of introspective attention, is that why is that unnatural? Consciousness is supposed to be something that we can directly be acquainted with, and I suppose the transparency point is, if so, then you’d expect something different than what Harman suggests that you will find or what my Challenge, in terms of process, would expect you to find. But I’ll take a look at the paper, thanks for the reference.
On the elephant, perhaps that will be made somewhat more clearer in the next post. My idea is that we do have a pretty good understanding about how attention works in helping us to latch onto perceptible features. So, in that respect, we have a handle on the process of attention, even if more work needs to be done. Given that, I try to draw an empirical consequence, but sorry, that’s the next post (soon!).
My answer about what information available to introspection is the transparency theorists: the same information available to perceptual attentional systems and, if there is a correlation between phenomenal properties and information about perceptible physical features, then information about the former too.
So, I think it does make sense to ask the questions I’m asking, but do let me know how you would ground the last question, which is an interesting one.
A pox on both houses, would be my answer. I think the literature has the problem upside down. I would argue that the best way to understand the peculiarities of introspection is as a kind of ‘natural anosognosia.’ We peer and peer into dark rooms convinced the lights are on.
Consider your test: “To the exercise again, and toggle attention back and forth between attending to [empirical] F and then to [phenomenal] phi. Is there really anything that you are clearly toggling except cognitive attention, namely moving between introspective and empirical thought on the basis of selective capacity?”
One of the things I always like to keep in mind when mulling these issues is the jaw-dropping difference in informational terms between Plato’s aviary, the metaphor he gives for memory, and our contemporary understanding of memory. The lesson, for me, is the way little or nothing in the way of internal structure is available to introspection. Why is this? Why is hazy homogeneity the default? Plato assumed memory was monolithic, one might suppose, simply because he lacked the information required to cognize its complexities. Memory intuitively strikes us as lacking internal structure because we simply lack the information, the ‘difference-making differences,’ to cognize any complications of structure. (And since we lack any information regarding this lack, what little we have is easily construed as ‘direct’ or ‘fundamental.’)
Essentially, you’re making an argument regarding the internal structure of introspective attention on the basis of introspection. You’re suggesting there’s no difference that introspection can discern, so why not assume it’s just the same capacity. But isn’t this precisely the mistake we’re prone to make given the informatic poverty of introspection? You could be saying, ‘I can’t discern whether anything might trip us up, so the floor must be clear of obstructions.’ The question remains whether you discern nothing because there are no obstructions, or because there’s no way to discern anything one way or another (say, because the room is dark).
But even if you pass on my pessimistic estimation, you still face the problem of heuristics. Heuristics solve problems by neglecting information, or what amounts to the same, by relying on the information structure of some specific problem-ecology. The question accordingly becomes, How domain-specific are the heuristics comprising our introspective capacities? Since the astronomical complexity of their target systems suggests ‘very,’ it becomes a very real question whether theoretical introspection or phenomenology isn’t largely if not entirely applying these metacognitive capacities out of school. My position falls out of my pessimistic answer, but the question strikes me as a pressing one for anyone pondering introspection.
I’m not making the argument that they are the same process (but this will be clarified in the second post, so sorry it’s not up yet).
My appeal to introspection is really to highlight how equivocal it is in this domain. I get the sense that some philosophers are confident that attention locks inward, that they can lock directly onto phenomenal character. I’m really just puzzled by that. For me, it’s more like transparency. But I’m not inferring from my introspection that there is the same process. If the post suggests that, it should be corrected.
The post ends with the suspicion that that transparency theorist will voice, given their introspection. But it isn’t to settle issues since I agree that introspection isn’t a very good way to discern the structure of inner processes although I would think it can give evidence.
Here’s how I intended the final paragraph to set up tomorrow’s post: introspection will in the end probably be equivocal. We don’t want to argue by dueling with our introspective reports…no end on that front. So, how can we move on?
I’m not sure I fully understand your invocation of heuristics, but to the extent that its a way of cutting through a lot of noise or dealing with gappy information etc., sure. This is tied to our discussion about attention, and the next post: I think we do understand quite a bit empirically about processes subserving attention. That’s the point I want to leverage in part 2.
But to dampen expectations: there are not going to be lots of fancy fireworks tomorrow, just a slight reorienting of the deployment of transparency, given the process-centered version. That might be too much navel-gazing among philosophers though to be generally interesting…
I think you were quite clear, now that I’ve reread the post! I think I was suffering the common ailment known as ‘selection for refutation’! But still, I don’t think my verbiage was entirely wasted. If you are interested in the way considerations of neglect and heuristics can impinge on the issue of introspection, you might check out: https://rsbakker.wordpress.com/2013/05/27/the-something-about-mary/ My whole approach is premised on the question of what we should expect the *brain* to be able to make of itself given its structure and resources. It’s remarkable how little attention this question has been given. And what I’m calling metacognitive neglect, if you give it some time to soak in, provides a pretty parsimonious way to explain our glaring inability to intuit internal structure. It leads to a particularly stringent eliminativism, however.
Thanks for this, Wayne! I don’t have time to respond properly now, but I’ll note that (i) I think that perceptual attention does embed — on my view, my awareness of redness is embedded (via perceptual attention) in a direct perceptual concept of redness, just as it is embedded via phenomenal attention) in a direct phenomenal concept of phenomenal redness; (ii) a lot of relevant issues (including a case for the ease of phenomenal attention and of switching) are discussed in my reply to Benj Hellie at https://consc.net/papers/contents.pdf.
Hi Dave! Thanks for the clarifications. I suspected that you would see the two forms of attention as on a par in respect of embedding, but thanks for making that clear. I will look at your reply to Benj (which you did recommend to me last year!) and perhaps bring out that case in the comments or an update!
For all that I say, other cases that are non-perceptual might provide stronger evidence for the direct model (e.g. moods, emotions, etc.). I take that point. But I hope post 2 will make clear why I focus on the perceptual case, so do stay tuned if you have time.
Hi Wayne. More great stuff!
I hate to bring this in, as I really don’t go for this intuition, but I thought at least some of the direct model folks are motivated by stuff like inverted and absent qualia intuitions. If you take this as something your theory needs to accept, then it seems a short step to direct acquaintance with something inner. (Though see Shoemaker for a complex attempt to avoid this.) Plus, there’s Mary. Anyway, if those are taken as placing theoretical demands on a theory–that is, if your theory has to allow that inversion and absence is a real possibility and Mary learns something new–then acquaintance seems a way to go. Plus, the relation is often called “sui generis,” leaving a lot of wiggle room about the details.
An interesting case is Papineau (2001), who goes for a phenomenal concepts view and has lots to say about the relation between “perceptual classification” and phenomenal concepts. (Not sure what he has to say about attention!)
Anyway, I’m all in favor of poxes here, but I feel like these intuitions are at play.
Here’s an attempt at a question/challenge for you. One might think that first we form concepts for perceptual qualities–the redness of a flower, the braying of a donkey, etc. And perhaps we can focus on those qualities in the world. Later (either evolutionarily or developmentally), we realize that we’re sometimes wrong about what’s out there in the world. And yet it still seems to us that the redness is present. We conclude that perception involves some sort of inner sensory quality. We now form concepts of these inner qualities, but we do not need to form wholly new ones. Rather, we “borrow” the perceptual concepts: red* is the inner equivalent of red. And it might inherit all of red’s relational connections: red* is more similar to purple* then to green*, and so on.
We now can introspect our sensory qualities as sensory qualities, but we do so in terms borrowed from the outer perceptual qualities. Then toggling might not bring a phenomenal difference,
This, so far, is very cognitive (indeed, it’s Rosenthal’s view, roughly). But imagine a phenomenal concepts version, perhaps of the nonreductive stripe. When we see green grass, we have a state representing the grass by using (in part) a green quale. Now we form a representation: “The experience ———” where the quale itself goes into that slot. We are aware of the same quality in both the perceptual case and the introspective case. In the first, the green quale is seen as a feature of the grass; in the second it’s seen as a feature of the experience. Again, we wouldn’t expect a phenomenal difference in toggling. All that happens is the quale switches from serving as the mode of presentation of the grass to a mode of presentation of itself. And you get to be a phenomenal inflationist (if you must).
Sorry, this got way too long! Anyway, I wonder what you think. Incidentally, I see attention as orthogonal to the toggling question, I guess.
Hi Josh
I can see that my post wasn’t as clear as I would have liked. I agree with your first point, these would be the theoretical and not introspection-based motivations for the direct model. I acknowledge those constraints, so part 2 of the post will raise a question: what if there were also an open empirical question that is relevant to these issues?
On toggling: my point is in the end that this isn’t going to settle much. I’m speaking from the perspective of what the transparency theorist is going to say, but I suspect that those who are firm with their phenomenology as being directly of phenomenal properties in a special and privileged way (and isn’t this the stuff of the history of our discussion of the topic?), will just differ. So the upshot is that really, no interesting purchase to be gained here by reflection. I’m ok with that, and did not intend to argue anything stronger.
I think your right that on some theories that are in line with the direct model, we won’t sense a difference in toggling. It’s a good point, but consistent with the inconclusiveness point that I meant to emphasize, that option (b) isn’t the way to motivate the direct model. So, I agree with your thoughts.
Question for you: why do you see attention as orthogonal to toggling? The toggling is precisely the redeployment of attention, as I’ve characterized it.
On toggling: It may just be definitional, but I was thinking that it was more a difference in how we think about our experiences, how we conceptualize (whatever that means!) them, rather than something attentional.
I was also thinking that you could toggle without attending, but that sounds odd when I type it. Maybe this is just another example (on my part) of muddy thinking about attention!
You’re right on toggling, in that how you describe it seems to me to be correct. There’s the cognitive component that is toggled (what I am calling cognitive attention, that is differently selective). The question is what does the toggling, and as the second part of your reply notes, it’s hard not to think that what effects the toggling is itself attention. But it’s easy to collapse things here, I agree. Attention is muddy!
Hi Wayne,
I’ve been really enjoying your series of Brains posts. Thanks for them.
I wanted to offer an answer to your titular question, “Transparency or Acquaintance?” and get your reaction to it. The answer is “neither.” The answer ties in quite a bit with higher-order thought approaches to consciousness, and so may end up connecting with issues raised earlier in the thread by Richard Brown and Josh Weisberg.
As you point out quite nicely, there are two theses one could label as “transparency theses,” one about the objects or targets of introspection (namely, that introspection shares its targets with perception) and another about the underlying processes (namely that introspection and perception involve a common process). I reject both of them.
Let’s look at the object/target-based transparency thesis first. Stating the basic thesis explicitly in terms of “attention” is quite inessential, as one could very well make the core claim at stake in much of the transparency literature in terms of “consciousness” and “awareness”. The claim can be made in terms of what one is conscious of or aware of in having conscious perceptions: one is conscious of what the perception is a perception of (the leaves on the tree), and never features of the perceptual state itself. This thesis fits nicely with first-order representational theories of consciousness. But it’s quite an ill fit with higher-order representational theories. The latter theories often get the ball rolling by claiming that conscious states are one’s own mental states that one is conscious of. So, contra transparency fans, one is conscious of the perception itself, and not just the leaves on the tree. For higher-order thought theorists one of the ways one can be conscious of something is by thinking about it and they ultimately hold that what makes a conscious perception conscious is having a thought about the perception. And clearly, I can think about a perception itself as opposed to the leaves on the tree–I can grasp in thought that the perception is one of my mental states and that the leaves on the tree aren’t. So, object/target-based transparency is false. Let’s turn now to process transparency.
Sensory perceptions themselves aren’t thoughts. And lots of sensory attention is most likely low-level non-cognitive stuff (for example, visual attentional capture by a flash of light). As Richard mentioned earlier in the thread, Rosenthal accounts for introspection in terms of third-order thoughts. This is to account for two kinds of conscious state: those that are introspectively conscious and those that are conscious without our introspecting them. Non-introspectively conscious perceptions are perceptions accompanied only by a second-order thought. Introspectively conscious perceptions involve the additional accompaniment of a third-order thought. Anyway, the point here is that the account of introspection is fully cognitive. Insofar as we want to regard introspecting one’s states as a kind of attention, (it does seem natural to say that in introspecting my state, I’m attending to it), then there’s an account of introspective attention in terms of a process that’s distinct from perceptual attention (the latter of which would include the non-cognitive processes of e.g. spatially selective attention of the leaf on the left as opposed to the leaf on the right). So, process transparency is false.
Let’s turn to acquaintance. I have difficulty even understanding what the view is supposed to be, but I guess that a core claim is one concerning directness. As I hear it, the directness claim is that awareness in the core cases is unmediated. Unmediated by what? Unmediated by inference, I guess. In seeing steam rise from the coffee, I infer that the coffee is hot, but my awareness that the coffee is hot is mediated by an inference based on my awareness of the steam. This, of course, doesn’t suffice to make my awareness of the steam direct, but it is at least more direct than my awareness of the heat in this example. Further, it may be the case that my awareness of the steam is only apparently direct. And this merely apparent directness can be explained by positing that it is mediated by an unconscious inference. If the evidence for so-called real directness is merely based on introspection (and who knows what else it could be based on) then that evidence can be explained away, since introspection involves only what is conscious and is silent on whether there is any unconscious mediation in play.
So, to wrap this up: introspective attention isn’t what the transparency people say, since different processes are implicated in perceptual and introspective attention and introspective attention isn’t a kind of direct awareness, since there’s no good reason for thinking that the relevant awareness is unmediated by inference.
Hi Pete:
Thanks for these detailed thoughts and kind words.
I do think that you gain something by switching to talk of attention and not just of talk of consciousness and awareness. You write, against the object transparency account:
“And clearly, I can think about a perception itself as opposed to the leaves on the tree–I can grasp in thought that the perception is one of my mental states and that the leaves on the tree aren’t. So, object/target-based transparency is false.”
I can’t imagine that the original theorists would deny the first point, but if thought is what is meant by being conscious of or aware of, namely thought about the mental state, then wouldn’t they (Harman e.g.) say that the switch in terminology has lost the essential point, which was attention? I think Stoljar’s article on diaphanousness is helpful here in distinguishing between cognitive attention and perceptual attention, and he notes, if I remember, that if Harman’s thesis is about cognitive attention, namely thought, then it is surely false too for the reasons you gave.
On the argument against the process account, let me just point to my second post which I will put up shortly and perhaps we can pick up the thread there, once you see what I want to say about the process version.
But I would raise a question for HOT theorists (I should revisit these views more carefully): What exactly is the thought about the experience? Is it a demonstrative thought about that experience or that feature of the experience? How does that thought get tied to the experience and its specific feature? I ask because I wonder whether more needs to be said about how attention informs higher order thoughts on HOT theories so as to tie the thought to the experience (or the thought to the thought) and if this lack of discussion of attention is something of a lacunae. But I would like to know more about that, so can I just ask for clarification?
I won’t say more about directness since it isn’t essential to my defending myself. Still, if one wanted a model, then one might invoke the idea of perception-based demonstrative thought that is mediated by attention as an exemplar of this sort of directness (ignoring philosophical attempts to first unpack directness). But set that aside (whew!).
Thanks for the remarks, Wayne. I look forward to your next post.
In the meantime, A couple of quick responses:
Re: what’s essential to transparency remarks. What strikes me as essential from Harman and the ensuing literature are the contents of consciousness: are they qualia, properties of external world objects, intentional contents…? The language of attention might just as well been replaced with Hume’s “when I enter most intimately with what I call myself…”. If it turns out that attention and consciousness are two totally separate things, then the transparency literature is not directly relevant to attention research. If they are the same thing, then there’s not much for you and I to dispute here.
Re:”how does that thought get tied to the experience…” There’s not been a lot of work on this specific question. Some thoughts, though: To insist that it’s demonstrative would seem to require the existence of the experience to be demonstrated, on pain of the higher order thought not being a complete thought. Some HOT theorists, notably Rosenthal, have insisted that the experience need not exist and there still can be a HOT that suffices for consciousness. It’s natural to think, then, that the structure of the HOT is quantified and descriptive. The non-existent experience is notional in such a case. The HOT is “tied” to the experience merely in the same way as “There is an x that is a horse and has a horn” is tied to its intentional content. Perhaps talk of “tying” is not particularly useful in such cases. “How do thoughts about unicorns relate to unicorns?” might not be a well-formed question.
Other HOT theorists, for instance Rocco Gennaro, insist on the existence of the HOTs targets, and that the HOT and what it’s about form a single state. A similar sort of move is made in the “same order” representational theory of Uriah Kriegel’s as I understand it. I don’t know off the top of my head whether either of those guys spell things out in terms of mental demonstratives.
FWIW, I’m inclined more toward Rosenthal’s way of thinking about this stuff. If what’s mainly wanted is an explanation of how things appear to us in consciousness, then it seems under motivated to posit a relation–demonstrative or whatever–to a real existent. If consciousness seems to us as such-and-such, it suffices to explain that to say that we think that it’s such-and-such. To my knowledge, the arguments against such an approach are conceivability considerations that I have a hard time taking seriously, e.g. so-and-so claims that they can conceive of babies and animals having consciousness without the posited cognitive capacities. But I don’t see that such alleged conceivings should impose a constraint on theorizing about consciousness any more than someone’s conceiving of a person walking on a rainbow should constrain the theory of light refraction.
Hi Pete:
On the first point, you might take my emphasis on the process-centered view of transparency to be a reaction to the treatment of transparency you note in the beginning of your reply: namely, object-centered transparency. I don’t think that is as fruitful of an approach, but once you focus on process-centered transparency, the issue of attention really is central. And, I do think attention is distinct from consciousness in that attention can be unconscious.
Anyway, maybe emphasis of process rather than object might be a useful contribution to the transparency debate, or so I hope.
On HOT stuff. In Part 2, I note the Many-Many Problem and the Many Phenomenal Properties problem. I have a latent worry about HOT, then, about how the thought gets connected to the lower level state. The worry is that attention will need to play a role if the HOT is to do its work. I need to think about this more.
Thanks, Wayne, for these very interesting questions! What follows is an attempt to remove the pox from the house of acquaintance.
You ask: “Why believe that there is any such thing as direct attention that allows for embedding?” My basis for this is essentially Russell’s test: that I can’t doubt that I’m now experiencing the phenomenology at issue. (I’d qualify his test just a bit, and say that my belief about my experience is relatively invulnerable to doubt.) The main claim about “directness”, in my view, is metaphysical rather than epistemic; but the reasons for thinking that our phenomenal judgments are sometimes directly tied to the phenomenal character of our experiences is epistemic.
So the question to ask your scientist friend is not “can you directly attend to phenomenal character?” Instead, the question would be whether your confidence that you’re now having a certain kind of experience (specified purely by phenomenal character) survives the supposition that you’re in certain skeptical scenarios. These scenarios might be introduced in stages. The first stage are the more familiar ones: brain in vat, etc. But to establish the “directness” claim, we need something more. The idea here is that, even if I suppose that the causal processes (etc.) in my brain are going completely haywire, I retain my confidence that this phenomenal property is now present in my experience.
Obviously, this brings us to Pete’s worries about unconscious inference. How can I tell, by introspection, whether this belief about my current experience is inferential? The key point is not that I’m unaware of a process of inference (this relates to Scott’s concerns). E.g., those who think that perceptual beliefs are inferential will usually grant that we’re unaware of inferential processes. Rather, the key point about inference is epistemic. Using a non-introspective case: if a belief (e.g. all walruses are walruses) survives certain doubts (e.g. doubting whether all walruses are mammals), that’s reason to think that my justification for the former belief doesn’t essentially depend on an inference from the latter. And the idea is that my belief about my own current phenomenal experience survives a suitably wide range of doubts, including doubts about whether the causal processes in my brain are operating normally. (I rush to say that I’m not claiming certainty about the results of such thought experiments.)
These epistemic considerations suggest, to me at least, that my phenomenal judgments can embed token phenomenal properties. So when I attend to my experiences, and form a judgment to the effect that I’m now experiencing this, my judgment is directly tied to its truthmaker, via the direct tie between the demonstrative “this” and the experience.
As Dave says: “The clearest cases of direct phenomenal concepts arise when a subject attends to the quality of an experience, and forms a concept wholly based on the attention to the quality, ‘taking up’ the quality into the concept.” ( Chalmers 2003 , 235)
I make the case for the acquaintance view, and show that it’s much less committal, epistemically, than ordinarily believed, in my 2012, available here.
A final thought: Josh’s idea, that toggling primarily involves varying how we conceptualize our experiences, is promising. In particular, I think that the order in which our concepts develop (start with the greenness of grass, eventually develop phenomenal concepts) will diverge from the explanatory or metaphysical order (phenomenal greenness is fundamental to our perception of green grass).
Thanks so much Brie, for your thoughts about this. As I said in the post, I didn’t do your view justice, and I’m grateful for you and Dave too, for taking the time to weigh in.
That quote from Dave is what I’m reacting to, in part. He says “the *clearest* cases of direct phenomenal concepts arise when a subject attends to the quality of an experience…” I don’t know, am I being daft when I say that it’s not clear to me at all that we have such concepts due to attention to the quality of experience interpreted as Dave intends that claim?
It’s the “clearest” that I”m leaning on here that make me think he might be taking option (b), but perhaps he just means the “basic” case given that we need to postulate attention of the relevant sort. Perhaps that would be what you would intend too.
As you make clear, a central motivation for you is epistemic. I want to agree that there is a kind of privileged epistemic status that one gains from introspection (though some commenters might doubt this). I’m not sure that you can’t be wrong even in cases like Dave’s clearest case or the one you bring up, but you might not be committed to that (as opposed to being commited to invoking embedding to explain those cases where our epistemic standing is appropriately high). I don’t have an argument against that, and as I note, the intuitions are compelling. I have ideas, but they are too half-baked to voice in this public forum (enough half-baked ideas of mine already out on the table!).
I’m after a weaker point in part 2, to spell out what the process-centered account comes to and then suggest that it might leverage in the other direction. It is, however, still an open question who is right: epistemic constraints pushing in one direction, empirical constraints pushing potentially in the other.