Dustin Stokes-University of Utah
dustin.stokes[@]utah.edu
The broad, probably overly ambitious, agenda of the book is to shift scientific and philosophical theories of perception away from one current orthodoxy. The orthodoxy I have in mind is modularity. The alternative I favor is malleability.
Anyone familiar with relevant literature knows that the controversy here concerns the modularity of perception, of the strong Fodorian variety. Few if any theorists debate that cognitive states like belief or desire or intention can influence one another, that one’s values can influence how one reasons, that a decision can and should be informed by one’s motivations, and so on. It should be only slightly more controversial that sensory processes and states influence one another, vision can affect audition and vice versa, and likewise for other senses. The continued debate concerns whether those cognitive states can influence, in some important way, perceptual experience. The strong modularist maintains that this influence does not, or rarely, occur. Perceptual processing is informationally encapsulated and therefore cognitively impenetrable.
As many readers will also know, challenges to modularity take the form of empirically grounded counterexample. A theorist identifies a case or set of studies where it appears that cognition is affecting perception and argues that the case is best explained as an instance of cognitive penetration, rather than a mere intra-perceptual effect, an effect on pre-perceptual attention, or an effect on post-perceptual cognitive states such as judgment or belief. There have been some promising putative counterexamples in recent years but suffice it to say that the debate has not really budged much.
The book takes two broad approaches to budge things. The first is the more familiar approach, and is the approach that I have taken in the past; provide empirically grounded counterexamples to the alleged informational encapsulation of perception. In this case, I at least attempt to shed some new light in a couple of ways. I suggest that extant attempts to define cognitive penetrability as such have largely failed and result in unhelpful theoretical cross-talk. I suggest that in place of a “real definition”, we characterize the phenomenon in terms of its consequences. The proposal I conclude with is disjunctive consequentialism, which says that “ψ is cognitive penetration if and only if ψ is a cognitive-perceptual relation, and ψ implies consequences for theory-ladenness or the epistemic role of perception or the behavioural role of perception or mental architecture” (106). I also argue that the modularists’ standard take on attention-mediated instances of cognitive influence on perception is misguided. There are plausible cases where cognition influences covert selective attentional mechanisms—such as feature based and object based attention; FBA and OBA—and whereby this in turn influences conscious perceptual experience. There are two ways to argue that such cases are instances of cognitive penetration. First, these mechanisms are part of perception rather than a gatekeeper between cognition and perception. Therefore, a cognitive influence on FBA/OBA is a direct cognitive influence on perception. This is framed in terms of biased competition models of perception. Second, one can argue that such influences amount to cognitive penetration by virtue of bearing important consequences for our epistemological and/or architectural theories of perception.
The second approach, and the one that I pivot to around the middle of the book, is to stop talking about modularity and cognitive penetration.[1] There, I said it. More carefully, I suggest that modularity does not deserve the status of default theory for the architecture of perception. This default position assumption is made by theorists on both sides of the debate, as evidenced by the pattern that pervades the cognitive penetrability literature. Opponents to modularity argue that a case violates some feature of modularity; proponents then deny that violation and maintain that modularity survives. Repeat. For this assumption to hold, for modularity to be the theory against which an alternative theory must position itself, that theory must be supported either by strong arguments or by superior explanatory power. Modularity enjoys neither. I defend this evaluation by, first, criticizing the clearest and most formidable arguments for informationally encapsulated perception. Those arguments center around the stability and reliability of perception, and none are strong enough to support the assumption in question. The rest of the book then makes the case that a malleable architecture better explains a large range of recent empirical studies and data.
Those studies concern perceptual expertise. A working characterization of ‘perceptual expertise’ underscores the following features: perceptual experts skillfully perform in a specific domain of training, their performance success is above a threshold set by the standards of that domain, and their performance non-trivially involves sensory perception. Such experts have been studied across a wide range of domain, from radiology to ornithology to fingerprint examination to elite athletics. Researchers use a range of behavioral, physiological (e.g. eye-tracking), and neurological measures (e.g. fMRI and EEG recordings). They also train subjects to become experts regarding “real-world” objects (e.g. cars or birds) and lab-created objects (e.g. “Greebles”), to better understand the acquisition and development of expertise. The best explanation of many of these phenomena is that the expertise is partly resident in the perceptual experiences of the expert, and those perceptual differences (by contrast to novices or the naïve) depend upon the richly cognitive training of the expert.
It is in this way that I support the claim that thinking affects perceiving (the TaP thesis), and in many cases thinking improves perceiving (the TiP thesis).[2] The TaP thesis divides into two architectural claims: 1) Some cases of perceptual expertise are genuinely perceptual, insofar as they involve differences in perceptual experience, and 2) those perceptual differences are sensitive to the cognitive learning specific to the domain of expertise. Of the first claim, studies on expertise show robust similarities with facial recognition (an undeniably perceptual phenomenon), where experts display standard behavioural and neural markers. Experts enjoy rapid and often “automatic”, successful performance and display significant differences in eye movement patterns. And they enjoy advantages in visual short term memory. This convergence of data (which I can only gesture at here) is best explained perceptually. By the same token, those perceptual differences depend upon the cognitive etiology of the expert, and this is the second architectural claim. Experts’ performance success, and persistence of that success, varies with fine grained learning of categories, and those changes are corroborated by lasting neural changes. Accordingly, mere “practice” or exposure to relevant stimuli is often insufficient for expert performance. And these skills tend not to “transfer” to similarly complex tasks in domains outside of the expert’s field. Perceptual experts are perceptual experts.
If one is compelled by these claims and their support, then in some sense the epistemology follows easily. Within a domain, perceptual experts perform reliably, rapidly, and with less distraction. They approach optimality. It is worth marking this as another (I hope) important shift in the book: an emphasis on cases of expertise is an emphasis on good-making cases of cognitive influence on perception (by contrast to the cases of error that populate the cognitive penetration literature). And so in Chapter 7 I argue that this successful performance, qua performance of the agent, is best understood in virtue-theoretic terms. This requires that perception can genuinely improve, and not merely as a matter of normal development or exposure to stimuli. Some experts acquire, through concept and category-rich cognitive training, through deliberate activity, a skill. The expert radiologist performs better visually because of what she has done, because of her actions, as a responsible epistemic agent. As a consequence of this training, her perceptual systems perform in exceptional ways within that domain. And those levels of performance near maximally satisfy the natural norms for perception, thus fulfilling the representational function of perception in optimal or near-optimal ways (in that domain). The important epistemic difference between this case and the cases of mere development or exposure is that the agent is herself responsible for the relevant etiology and, accordingly, for the perceptual improvement. The epistemic virtue is therefore attributable to the agent herself (not just, say, her visual system). In cases of expertise, thinking thus improves perceiving.
The TaP and TiP theses are the descriptive and normative components of a theory of how thought may affect perception, of how the mind is richly malleable. Important consequences follow. But since I haven’t the space to expound at length about them, I’ll offer (more?) provocation in hopes to compel a few readers. Accuracy is but one facet of perceptual success and thus just one possible determinant of perceptual content. Perceptual success can also involve increased sensitivity to gestalts, patterns, and feature types, achieved rapidly and efficiently, integrated with action, and with less distraction. Success along these measures can vary from domain to domain and so it follows that perceptual content is not determined in a purely mind-independent Objective* way, but instead in an inter-subjective, objective way. Determinants of content thus include facts about the environment, but also facts about the perceiver’s epistemic community. This is to accept theory-ladenness, both in science and in the ordinary course, and with that comes important risks and possible vices. One such vice that is discussed is implicit bias and the “cross-race effect” in facial recognition. It is also to accept rich perceptual content but without admission of kinds into that content. Enhanced perceptual sensitivity as enjoyed by experts– to patterns, gestalts, and organizational features – is to enjoy rich perceptual content. Importantly, this lesson is partly learned by considering cases of perceiving aesthetic properties: The ballet instructor sees not only the colours, edges, shapes, and motion of her pupils but also how those features are organized in ways that are balanced or serene or graceful. Some aesthetic experts are perceptual experts. Finally (for now), the phenomenon is general. Although the experts described here are remarkably accomplished, they are not super-humans. All humans are habit forming and many of those habits involve perception in non-trivial ways. Therefore, genuine perceptual expertise is, I suggest, a pervasive phenomenon. We are all of us potentially perceptual experts and in a variety of contexts. To accept this kind of malleability is, I think, to better understand ourselves and our place in the world.
That, in any case, is my attempt to distill some of the main themes of the book. I want to thank my excellent commentators in advance—Becko Copenhaver, Jonna Vance, and John Zeimbekis. I look forward to their feedback and learning from their criticism. Thank you also to Dan Burnston for pitching the idea for this symposium and to he and others at the Brains Blog for seeing it through.
[1] A distinct reason to abandon talk of the latter is that it is, let’s be clear, an ugly phrase.
[2] I know I know…nerdy acronyms.
Thanks for this stimulating summary, Dustin. I look forward to the commentators’ responses.
Quick question about the ‘T’ in ‘TaP’. Does “thinking” have to be conceptual?
I follow a conceptually deterministic procedure to perceptually determine whether a shape is convex. My ability to do this is dependent on the concept and doesn’t seem to require a change in perception. On the other hand, I learn to distinguish Gothic from neo-Gothic buildings by learning (or being taught) to look for non-conceptually characteristic features. This requires (or will bring about) a change in perception, but does it require thinking?
I wonder if this distinction is just too binary to capture your intentions.
Many thanks for this comment, Mohan. The cases I have in mind in the book typically do involve explanatory appeal to some kind of conceptual learning: facts about radiology diagnostics, different levels of bird or car categorization, valid contexts of play in sports. In these cases, I argue, the perceptual changes causally depend upon the concept-rich learning specific to that domain. And I appeal to studies of these cases to motivate both the TaP thesis and the TiP thesis. In other words, I do think that some of these kinds of cases are conceptual in a robust sense and support the theses interpreted conceptually. All of that said, I have no objection to another version of the theses which needn’t be interpreted conceptually, and your Gothic/neo-Gothic example sounds entirely plausible to me on that score. And indeed, I want to leave space for perceptual learning of this sort that does not obviously require explanatory appeal to concepts, while it also clearly involves active practice and engagement on the part of the perceiver. I think this topic will surface later in the week when I reply to some of Becko’s nice commentary on perceptual learning versus perceptual expertise. Thanks again.
I’m looking forward to reading the book! It sounds fascinating and the main issues and arguments are laid out really nicely here. One question I have is with respect to the second component of the TaP thesis, “2) those perceptual differences are sensitive to the cognitive learning specific to the domain of expertise.” If you abandon the language of cognitive penetration in this part of the book, then I am curious how you are thinking of ‘sensitivity’.
It seems the sort of sensitivity you require for TaP to be a substantive thesis is precisely along the lines of cognitive penetrability because in a weaker sense, most everyone would grant that perceptual learning is influenced by cognition in some way. For example, in the case of birdwatching, learning the name for a kind of bird might incite someone’s interest in the domain, causing them to go out and actively look for exemplars to identify, which then puts them on the path of becoming a perceptual expert. But this sort of sensitivity seems to fall short of the variety that I take it you are interested in.
Alternately I wonder whether in many cases there could be a common underlying cause (such as interest in a domain) that promotes cognitive and perceptual learning at roughly the same time, rather than explaining perceptual learning as being guided by cognitive learning. To give an example from procedural learning, the baseball professional’s ‘know-how’ doesn’t automatically qualify as a case of cognition affecting motor ability in any thick sense just because her interests caused her to learn a lot about baseball and spend a lot of time practicing. Her baseball know-how could have developed independently from her cognitive learning about baseball, but the co-occurrence could make it look like one is causing the other.
Maybe what is behind the TaP2 thesis is that the categories that the experts learn must be learned by cognitive means. Is this a fair interpretation of your statement that “Experts’ performance success, and persistence of that success, varies with fine grained learning of categories, and those changes are corroborated by lasting neural changes”? If so then I am wondering why the categories couldn’t themselves be perceptual categories. That is, how can we rule out the alternate explanation that the perceptual system itself is what learns the relevant categories and subsequently sorts objects into these categories? Thanks for your thoughts on these matters.
Many thanks for this comment, Madeleine. You raise an important concern for me (actually a handful: but I’m going to focus on your comment about whether an interesting version of TiP requires some commitment to cognitive penetration). There is a lot to say here, but here is at least part of my reply. First, although I do spend a good portion of the book criticizing modularity and making the case for cognitive penetrability, I really do attempt to shift emphasis away from both (around the middle of the book). Perhaps this is overly heavy handed as a rhetorical technique but for many reasons I feel that those debates have gone a bit stagnant (and for that matter I may have contributed to that situation). But there is also a more principled reason for attempting to defend the TiP thesis without commitment to cognitive penetration. I think there are a variety of phenomena in the vicinity, and they will likely require a plurality of explanation types. Here is part of that analysis from the book (these are excerpts from the book; but I’ve tried to contextualize them here).
“The perceptual effects described [in Section 6.3] non-trivially depend upon cognitive learning: information about categories, diagnostic detail, goals and tasks within that domain. It should be emphasized here, and will be further clarified and justified later, that this claim is neutral with respect to the particular means or mechanism whereby expert learning or cognition affects visual perception of experts. Instead, [the relevant claim], architectural claim 2 embraces an explanatory pluralism, and that pluralism opposes a disjunction of possible claims made by [the] sceptic”.
That sceptic “denies that cognition or higher-level mental content can have a significant effect on vision (and perhaps perception generally). For different theorists, this will come to different claims. Some will claim that cognition does not penetrate perceptual experience or perceptual processing. Some will claim that there are no top-down effects on visual processing (synchronically or diachronically). Others will claim that there is no top-down modulation of rapid visual attention or visual fixation or saccades.
The conclusion of this section, [architectural claim 2,] can be interpreted as denying each of these sceptical claims but without committing to a single means by which relevant cognitive-perceptual effects are achieved” (162-3).
A few pages later, I offer the following,
“For all that has been said, the architectural claim (or conjunction of claims) is a broad one: Many instances of perceptual expertise, such as those just discussed in detail, involve cognitively sensitive perceptual changes and differences. But no claims have been made about what more precise structure or mechanism is involved across the array of cases. Thus, some cases of expertise may involve cognitive penetration of perception, some (perhaps non-exclusively) may involve high-level perceptual content, others differences in mechanisms of selective attention. This pluralism is deliberate. There is little reason to think that all instances of perceptual expertise take the same structure. While there is a convergence of behavioural and neural-physiological evidence across many domains of expertise, there are also important differences across those domains. Some such domains require visual search as the dominant task, others involve object categorization, others still some kind of perceptual matching or reidentification of an individual stimulus. It is plau- sible, then, that the relevant cognitive-perceptual effects might be achieved differently across different domains. The claim meant to take broad scope across these varied domains, argued here and contrary to a disjunction of claims made by sceptic 2 (or distinct sceptics), is simply that many cases of perceptual expertise are both richly perceptual and importantly cognitively sensitive” (166).
Now I see the (or one) possible worry: this may be my trying to have it both ways. But I wanted to give an analysis of a range of good-making cases of cognitive influence on perception, of perceptual expertise, without getting embroiled in various definitional issues about cognitive penetration, debates about diachronic versus synchronic effects, early versus late stages in visual processing, informational encapsulation, and so on. And here I confess that I really do try to have it both ways, since I end up spilling some ink on most of these issues after all. I just didn’t want the value of the analysis to stand or fall with whether it succeeds as a counterexample to modularity. I wanted to try to defend a broad kind of malleability of the mind on its own explanatory merits. This dovetails importantly with some of the critical analysis earlier in the book, where I attempt to undermine the default theory assumption (as I describe above in the introduction).
That, in any case, is what I am inclined to say for now. But your concerns are well taken. I’m sure we’ll have more opportunity to talk about these issues, including your own work on related topics, in the near future! Thanks again.
Hi Dustin, I’m looking forward to reading the book and the discussion.
I wanted to ask you about the idea that modularity does/doesn’t deserve to the be our default hypothesis.
Suppose (as many do) that modularity/encapsulation can be identified with, or at least seen to imply, cognitive impenetrability. You note that, with this assumption in view, critics of modularity tend to try and find compelling cases of cog pen, and proponents of modularity try to deny these, the (shared) assumption being that insofar as these cases don’t exist modularity wins.
I’m wondering why you think this assumption is problematic. It seems very plausible to me that, if cognitive penetration were never to occur, this would strongly suggest some kind of architectural restriction on the flow of information between thought and perception. Do you disagree with this, or do you instead think it would imply some kind of architectural restriction that falls short of full-blown cognitive impenetrability? If so, how?
I realize that (in reality) you think there are lots of cases of cognitive penetration, and I’m excited to look into the cases you describe more closely. But I’m just trying to get clearer on how you see the dialectic and precisely where you see the burden lying for modularity enthusiasts.
Many thanks for your comment, Sam. I think (or at least hope) that some of my reply to Madeleine (above) will be relevant here. But the short of my reasoning here is this. I think cognitive penetrability has become the benchmark for any interesting cognitive influence on perception, and that this is a consequence of its incompatibility with strong versions of modularity. As a result, the value of analyses and explanations of possible cognitive influence on perception hinges on whether they serve as counterexamples to modularity. This betrays what I’ve called the default theory assumption. And I simply don’t think that the “classic” arguments for modularity support the assumption. Accordingly I spend the bulk of Ch. 2 of the book criticizing those arguments. And I don’t think that modularity, as a computational posit, is sufficiently explanatorily powerful to warrant being the default, once we consider a range of empirical evidence. Put differently, I find this assumption limiting in terms of the kinds of studies and phenomena that philosophers of mind address and how they analyze them.
Dustin, great stuff!!! I have a very basic question: Is modularity really the default in philosophy these days? It was in the 80s, but then again, we used to listen to Duran Duran in the 80s… And not even then was it the default in psychology (where now it is in extreme minority if it is even statistically significant). I can think of maybe half a dozen philosophers, mainly young, mainly vaguely in the grip of some degree of Rutgers influence, many of them my good friends, most of them brilliant. But definitely not the mainstream. And even they are not modularist in the sense you’re interested in, but rather, for some inexplicable reason, trying to squint so hard as to give some kind of over-charitable reading of classic modularist claims that would be empirically plausible. I do think that your line is a novel and cool direction in an old debate but I do wonder whether it’s best to frame it this way.
Many thanks for your comment, Bence. I’m hoping that some of my reply to Sam (just above) addresses your question. I think that modularity does remain the default (or an orthodox default, even if not the only one) as an architecture of the mind, in particular as pertains to perception and its relation to other mental processes. Again, I think whether a phenomenon is explained as cognitive penetration has become something of the benchmark for whether it’s an important or interesting cognitive influence on perception. And it’s that kind of assumption, grounded in strong Fodorian modularity, that I resist. And I agree with what you intimate: this impression can really vary with the context. When I’m spending time with some folks (say, you), then I think the debate/s has moved on and in ways I endorse. But when I hang out with others (presumably some of the very same folks you indicate in your comment), then I think that default assumption is very much alive. Finally, it is importantly not the default in philosophy of perception more broadly. This is sometimes for the somewhat trivial reason that there are just different explananda in our field. For instance, you don’t hear intentionalists and disjunctivists invoking terms of modularity in their debates about the structure or metaphysics of perception.
Wait, and are you saying that you don’t listen to Duran Duran anymore, Bence?!