The final part, though I might add something about some issues that have arisen in comments. Anyway, here’s the idea to be unpacked:
Introspective and perceptual attention rely in the same way on visual experience, namely the coupling of vision to cognition
Attention based Coupling
Let’s say something about coupling (I borrow the term from Alan Allport). Talk of coupling emphasizes a crucial role of attention. Elsewhere, I have written about the Many-Many Problem, the idea that to perform perception-guided behavior, we need to select some aspect of what we perceive to inform action. This selection for action, I claim, is a form of attention. In general, we can think of fixing beliefs about the external or internal world as often a goal-directed form of behavior.
Note that there is a Many Phenomenal Properties problem. A visual experience will exemplify many phenomenal features. Thus, to think about a specific phenomenal feature rather than others on the basis of introspection, one must select the feature, namely attend to it. We can then, for a given visual experience at a time, draw a space of possible introspective thoughts, each coupled with a specific phenomenal feature of the experience in question. Each potential introspective thought would be grounded in appropriate introspective attention to the phenomenal feature. The actual entertaining of a specific thought results from actually attending to the feature: so we have two modes of attention, the selectivity for cognitive attention, and cognitive attention (the thought) itself. In this way, visual experience is coupled, by introspective attention, to the relevant thought. For the direct theorist, so far, so good.
Let us now turn to visual attention that informs empirical thought. Here, we have the same structure as we saw in introspective thought, a space of experience based possible thoughts where attention functions to couple an input to thought: to think about the world on the basis of vision, we must exhibit selectivity of visible features of the world which informs our thoughts. Attention must again give rise to a coupling from visible features to thoughts about those features.
The upshot: in the space of possibilities for empirical and introspective thought, we have vision playing the same role. Visual experience is the basis of the vision-cognition coupling effected by attention, whether perceptual or introspective. You might call it the input, but that doesn’t sound the right note, so I won’t do so.
Both direct and transparency models of introspective attention can (and should) agree to this shared structure. Direct theorists, of course, will reject object-centered transparency, for they take the input to introspective attention to be a phenomenal feature whereas they take the input to perceptual attention to be the physical feature. The object-centered transparency theorist seems to say that the object of both forms of attention is the physical feature (but what of hallucinations!).
Shift now to the process-centered version of transparency, and focus on the shared structure noted above. Here, there is some empirical purchase on the issue and this is what I want to bring out. The process of visual attention is one of the most intensely studied of psychological phenomena, with much important work being done since the 1960s. While there are many controversies and confusions in this domain, I think attention is one of the better understood psychological capacities. We have lots of data that flesh out a picture of the mechanisms of attention. Thus, if we turn to the vision-cognition coupling, we have some good ideas of the mechanisms that underwrite it, both at the neural and psychological levels.
I don’t want to oversell our understanding of attention, for there is much work yet to be done. But let’s get a sense of the overall shape of vision-cognition coupling. What seems relevant is the selection of contents regarding the attended physical feature, F, such that those contents come to selectively inform cognition. At the level of mechanisms, the idea then is that the brain carries information about F (in whatever technical sense of “information” is of most use to neuroscience) where this information is selected to inform cognitive systems. This selection can be effected by a variety of mechanisms that have been discussed in the empirical literature: biased competition (up- and down-regulating signals when presented with multiple stimuli), receptive field remapping (filtering), gain modulation (amplification of signal), divisive normalization and so forth (I discuss this in a chapter on the neuroscience of attention in my forthcoming book).
Ok, so assume that we can identify this mechanism (or family of mechanisms) of the vision-cognition coupling which then explains how visual information is channeled for cognition. Call the mechanism of attentional selectivity N. N tells us how the relevant visual information about F is channeled so as to inform empirical thoughts about F.
Now: might it not be the same mechanism N that is deployed in introspective attention?
An Empirical Upshot and its (possibly) interesting Significance
I don’t know what the answer to the last empirical question is, but consider the affirmative answer. Then it seems that the transparency theorist has gained something by emphasizing the same-process version of transparency. For on the affirmative answer, the information used by N also underwrites the vision-cognition coupling that leads to introspective thought. That is, attention-based empirical thought and attention-based introspective thought share the same realization in terms of the visual basis that is coupled to thought. The first point then is that this mechanism is in a certain context C sufficient for attention-based empirical. That is, it is the basis for perceptual attention to features in the world that ground our thoughts about those features. The second point: in that same context C save with a shift in goals from fixing empirical thought to fixing introspective thought, N will be sufficient for attention-based introspective thought. Accordingly, the third point, there might be no additional mechanism of attentional selection beyond this shared basis in N. Specifically, there isn’t a form of attention that directly targets phenomenal properties. Rather, the form of attention at issue is one that deploys the mechanisms of vision-cognition coupling as exemplified in the perceptual attention case. We could talk about perceptual attention as the common element, but there are reasons I want to avoid this and simply speak about the same vision-cognition coupling.
The transparency theorist can leverage the empirical possibility we just noted to ask of the direct theorist what the realization of their model would be, a form of attention that targets a phenomenal property directly such that the property is embedded in introspective thought. This is pressing because if attention is something like a selective capacity, then N characterizes (realizes) this functional role. Accordingly, one can’t just talk about attention without some take on the realization, at least in principle. If the only basis for postulating such a distinctive capacity for attention is something like introspection (it’s obvious we can directly attend to consciousness), then I am deeply suspicious (though Dave Chalmers has some cases that might work, I will try to comment on them if time).
The direct model has another route. In the last post, I noted that the direct model can also be motivated by theoretical constraints from other areas of philosophy such as epistemology. In that sense, the direct model of introspective attention treats it as a theoretical postulate. For we seem to need something like introspective attention of the direct sort so that we can satisfy the various conditions that we impose on introspection: (a) its generating a distinctive way of thinking about the mind; (b) its privileged epistemic standing; (c) its distinctive first-personal aspect, and so on. [Josh Weisberg notes some other options in his comment in Part 1].
I don’t have answers to those admittedly weighty considerations. But here’s the crucial point that I want to emphasize. There are powerful epistemological constraints on the one hand that push us to certain views about introspective attention. These constraints have shaped our discussion for some time now [update: please see Brie Gertler’s helpful comment in part 1]. But if there is an open empirical question of the sort I noted above about N, then it is an empirical possibility that introspective attention is in a sense the same process as perceptual attention in that it relies on the vision-cognition coupling that N realizes, the functional role of perceptual attention. If that were true, then we can push in the other direction: might this now change how we should theorize about the epistemological and cognitive assumptions many philosophers have made regarding introspection and phenomenal concepts? For we would need to explain those phenomena without the idea of directly attending to phenomenal properties, that being a distinct process than the vision-action coupling we have noted.
Granting all that, then there is a way in which some of the central issues regarding introspection and consciousness have an empirical component, or at least that in principle, investigation of the underlying mechanisms might b. If I’m right, then while I can’t give a definitive answer whether N underwrites introspective attention, I can say that it behooves us to flesh out the transparency model in its process-based version and to see what its potential philosophical significance is, how it would reframe questions about the properties and structure of introspection and of phenomenal concepts. After all, for all we know, it might be true. This is the theoretically sound approach to a crucial issue both in philosophy and psychology: the nature of introspection and its epistemic status. It is clear, I hope, that reflection on attention plays a crucial role.
Okay, I wasn’t so far off. In my own case, the very reason why I find your attention toggling exercise unconvincing (because the absence of intuiting any distinct introspective processes likely speaks to neglect, the fact that introspection lacks the information required to make the distinction) informs my hunch that something like the coupling you describe is indeed the case. In other words, I think you might be right, but for quite different reasons.
I also agree with you that one needs to undermine (a), (b), and (c) for this hunch to be compelling.
Again, I want to underscore simply how curious this situation is: We intuitively take ourselves to be doing something ‘obvious’ when we introspectively attend, but without any obvious intuition of the dynamics and structure of whatever that ‘something’ is. So to repeat my question from earlier: Why is fuzzy homogeneity the default?
Let’s call this the ‘fuzzy homogeneity’ problem.
It’s always struck me that ‘fuzzy homogeneity’ is also the default for the scarcity of information in empirical cognition as well. Absent the *availability* of difference-making differences, we seem to regularly assume homogeneity (as when we mistake ants on the sidewalk for spilled paint, for instance). Isn’t really so far off to describe scientific progress as the gradual complication of apparent simples? If it were the case that introspective attention turned on, or coupled with (the ‘coupling’ metaphor suggests something too clean, I think), empirical cognitive capacities, then we should expect introspective fuzzy homogeneity to pertain to instances of metacognitive information scarcity in a manner similar to the way low light conditions, say, ‘homogenize’ our visual field, but – and this is the crucial bit – without the information required to realize that information scarcity is an issue. This essentially Plato’s insight in his cave allegory: If we were raised in a cage in a dark room, we would have no access to the information required to sort the ‘fuzzy homogeneity’ pertaining to incapacity from what was the case.
This is the cornerstone of my entire upside down approach: the question of realization is all important, because short of answering it, the introspectionist has no way of advancing *any* credible claim regarding the structure and activity of introspection. So for me, the questions you’re working toward are:
1) What information is accessed in introspection?
2) What cognitive resources are deployed in introspection?
3) Are (1) and (2) adequate to the kinds of questions we are asking of introspection?
And it just so happens that there’s one particular way of answering these questions that allows us to *explain away* (a), (b), and (c). But that’s a much bigger story…
Hi Scott:
I think in broad outlines, I agree with the central points you are making about introspection.
Your first question, at the end, is really important to me now. The process view suggests the “tracks” by which information is conveyed, but of course, the introspective thought processes have to be abstracting different information than the empirical thought processes, else they would be the same process (i.e. same output). The point to work out is that the same coupling, namely the visual experience modulated by attention, is the “common signal” for introspection and empirical thought. But this really has to be worked out.
Information here is the crucial notion, and its connection to contents. This is a huge topic, controversial. In respect of introspection, some attempts by Dretske and more detail from Aydede and Guzeldere in a Nous paper some years ago are notable. I’m not endorsing those views, but the ideas of information are taken seriously. This is where my thoughts really come to an end though! It’s for further work which I haven’t done.
I take information as an unexplained explainer: differences making systematic differences. Before wrestling in the semantic information mire, at least consider that our intuitions regarding content themselves also derive from metacognitive processes. On the one hand you have the profound ‘curse of dimensionality’ that we know (short of spooky spontaneous autocognition) the brains faces attempting to cognize its own operations, and on the other hand you have the missing causal dimension in intentional concepts that no natural explanation can apparently bridge. Could be a coincidence. But it is striking, and it raises the possibility that much philosophy of mind turns on chronic heuristic missapplications. Beginning with me, no doubt!
Of course, all I have ended with in this post is not a claim, but a hypothesis, and (OK), a conditional claim: if the hypothesis is empirically verified (how in the world might we do this?), then certain consequences follow. Given that the hypothesis is on the table, we should explore its potential consequences.
Hi Wayne,
This is really stimulating stuff. Thanks again for your Brains posts.
I intend the following to pick up somewhat on our thread on Part 1, but we’ll see how it goes.
I’m not quite following your line of thought on the Many Phenomenal Properties problem and the general structure of what you see as the solution to it. As I see it, you’re recommending that there are three things here: the phenomenal experience, attention, and the introspective thought. I don’t see that you need the first two, and I especially don’t see that you need the second one.
Consider some general remarks about thought. Right now I’m thinking that Richard Brown is a pretty good drummer. (BTW, I’m pretty sure he’s almost 20 miles from where I am right now, and I haven’t played music with him since December.) Of course, Richard Brown has lots of properties, and his being a drummer is just one of them. Echoing your Many Phenomenal Properties question, we might ask how it is that I’m able to think about one of those properties and not some other. But I don’t see that we need some specific capacity–attention–that’s distinct from thought, to explain it. It’s the dynamics of my train of thought, whatever those are, that leads me to think about that feature of Richard. I might even think of properties of Richard that he doesn’t even have, like that he’s a millionaire. I don’t see that there needs to be something that literally connects me to Richard and his properties that “selects” his properties for me to think about. Whatever accounts for me thinking about his drumming should be the same mechanism that allows me to think about his (nonexistent) vast wealth, and the latter need not involve some literal connection to a property of Richard.
Now, you might think that this lack of connection to Richard only applies insofar as we’re focusing on a particular sort of case, but if we change to a different sort of case, a connection becomes necessary. If I’m currently perceiving Richard, then, plausibly, its a requirement on my perceiving him that I’m in current causal contact with him. That’s what makes certain of my Richard-directed mental states count as perceptions of him as opposed to merely thinking of him. Nonetheless, I can think of him in inaccurate ways while perceiving him. I perceive his swanky new tie and think “Aha, a millionaire.” Again, I don’t see there needs to be a connecting mechanism distinct from my thoughts to account for my subsequent thoughts, since plausibly, the same explanation should suffice for the true thoughts and the false ones.
Next you might motivate the positing of a connection mechanism by appealing to the fact that I’m generally reliable at thinking true thoughts upon perceptual presentations of Richard. Not to pat myself on the back, but I generally come to think of him as he really is. Therefore, there needs to be some mechanism that explains this by mediating between actual properties of Richard and my non-miraculously true thoughts about him.
That’s not a bad move, but the trick, then, will be to establish a similar point about introspection of so-called phenomenal properties. And now all sorts of vexing special problems arrive on the scene. One involves Schwitzgebel-type points about whether introspection actually is generally reliable at delivering true thoughts. Another involves Rosenthal-type points about whether you need anything other than the thoughts themselves to explain so-called phenomenal properties, those properties in virtue of which “there’s something it’s like” (whatever that means) to have conscious states.
So anyway, I intend this as a brief in favor of a third model for introspective attention, one that’s neither the transparency view nor the acquaintance view. On one reading of the model, it’s eliminative about selective attention, if that’s supposed to be something separate from the dynamics of thought. On another reading, it’s reducing selective attention to whatever it is that generally accounts for me thinking about whatever it is that I’m thinking about right now.
I am generally sympathetic to Pete’s remarks (especially about my drumming ability 😉 In addition to the line of argument that he makes I would also suggest that we have some evidence that attention is neither necessary nor sufficient for consciousness which suggests that attention isn’t necessary to form the kind of higher-order thoughts that HOT theories of consciousness invoke, and if so no worry about coupling.
However, I don’t see why we can’t have a mixed view here. One could hold that selective attention is one of the ways that cause the appropriate HOTs. Is there an argument against this kind of view?
Hi Pete, Richard:
Let me reply to both of you. The coupling was raised because I’m fixated on a specific view about introspection, namely that it is something akin to a demonstrative or involves embedding. I have to respond to Brie’s comment in Part 1 which might be of interest to you too. So, I don’t have a general requirement that to have a thought about X, one needs to attend to X. Though I do think that to have a perception-based demonstrative thought about X, one needs to attend to X.
Richard’s drumming is legendary, but I can only think about it descriptively. Pete and Richard can think about it in a more intimate way. But do we think about phenomenal properties descriptively…well, actually I wonder whether we might in fact do just that!
I’d like to talk more about HOT, but I think the issue with respect to HOT theories needs a bit more detail from the HOT perspective. What is the specific content of the relevant HO thought that makes one state rather than another conscious? Depending on how you spell it out, then I will either agree with you that attention is not required or that it is. And Richard is right, there might be a mixed view.
I should add: if there is an empirical question whether N is present in introspection in general, it should be an issue for any theory of introspection. Maybe HOT’s theory of introspection doesn’t imply attention. Good, so, then it would be of interest to HOT to see if the science of attention, properly applied to this area, might reveal whether there is a shared N that is causally implicated in introspection.
Thanks, Wayne. That’s helpful. I thought the general structure of the argument was a disjunctive syllogism: either acquaintance or transparency; not acquaintance; therefore, transparency. I was questioning the truth of the disjunct by offering a third alternative. That alternative, as I sketched it, does depend on thinking of introspective thought as descriptive (btw, I think all thought’s descriptive). Anyway, your recent remark helps clarify how you intend to limit your argument’s audience to demonstro-philes.