Dustin Stokes is associate Professor of Philosophy at the University of Utah. His research includes work on perception, imagination, and creative thought and behaviour. His visit to the Institut Jean-Nicod as Visiting Director of Studies at EHESS is an opportunity for the scientific community to discover or learn more about his work through a series of four lectures.
EHESS Director of Studies and member of the Institut Jean Nicod, Jérôme Dokic is a philosopher of mind and cognitive science working on various topics including indexicality and situated cognition, perception, memory and imagination. He introduces us to the visiting professor through an interview questioning his work on how thinking influences perceiving and and its interdisciplinary working methods including philosophy and cognitive science
J.D : Can you explain in an informal and broadly accessible way your view that thinking influences perceiving and the consequences of this view for our conception of the boundaries between cognition and perception? For instance, why does this view not entail that there is no interesting difference after all between thinking and perceiving?
D.S : Yes, of course. My view says that thinking affects perceiving and, in some cases, thinking improves perceiving (perhaps unfortunately, these are named the TaP and TiP theses in the book my book—Thinking and Perceiving, Routledge, 2021). So I suppose the first step is to clarify why these are not trivial theses. There are uncontroversial ways in which thinking both affects and improves how we perceive the world through the senses. If you believe that there is an optimal way to attend some event or act on some feature of a thing, and have the motivation to do so, and then act on this decision, you will change how you see. The best way to be certain that the fish is cooked is to look and see, as we say. You touch it with a fork and if you see it flake in the right way you determine that it is cooked. Your decision and resulting action in this case affected what you saw and improved it in the context (by contrast to a case where you decide to not look at all). Dangerously, this phenomenon can go the opposite direction and have important social-political consequences (especially where I’m from). If you decidedly do *not* look where the evidence lies, thought is affecting (and not improving) perception by sustaining ignorance. And you can’t justify this ignorance by saying you didn’t see it: you were deliberately looking the other way.
While these are interesting and important phenomena, my interest is in phenoma where thought influences perception more directly. I am interested in cases where thought changes perception and not just because a perceiver has decided (or not) to look at something. In the context of debates about the so-called “cognitive penetration of perception”, theorists (on both sides) tend to identify cases where one’s cognitive states somehow hinder perceptual representation: one suffers an illusion or misperceives as a result of what one believes or desires. I’ve been party to this debate and think there are plausible cases. But in my book—Thinking and Perceiving (Routledge, 2021)—I focus on perceptual expertise, as studied in psychology and neuroscience. I’m interested in these phenomena for a number of reasons, not least because philosophers of perception have paid little attention to them and they seem to be plausible good-making cases of cognitive influence on perception. The expert radiologist or footballer or forensics examiner appears to better visually perceive stimuli within her domain of expertise, and as a non-trivial result of her conceptually rich training within that domain.
This means that the very look of things—the phenomenal experience of the radiogram or the penalty kick—can be different for different perceivers (say the expert vs. the novice). But this does not imply that the distinction between thought and perception dissolves entirely. In fact, notice that the very discussion presupposes some such distinction. In the book I offer a number of ways that we can distinguish cognitive states and processes from perceptual states and processes. The first do not, in any one instance, rely upon relevant sensory stimulation nor an appropriate external stimulus: one can believe that the sun will rise tomorrow while one is in a pitch black room or imagine the flying unicorn while at the tennis match. And these states, by contrast to visual and other perceptual experiences, lack the same rich phenomenology. Visually perceiving the sunrise requires retinal stimulation, and the sunrise (at least when vision is functioning properly), and it will be rich in phenomenal character. What it’s like to have such an experience will be rich from the first personal perspective (by contrast to a belief about roughly the same thing). These are dimensions, perhaps among others, along which we can keep the cognition-perception distinction. It’s just that on my view the first can influence the second in interesting ways.
Finally, and perhaps this just reiterates what I just said, this doesn’t imply that the two are entirely continuous. It doesn’t imply that cognition is *always* influencing perception. I think opponent views like modularity sometimes characterize historical allies (of mine) in this way. They assume that in maintaining *some* cognitive influence on perception, one must maintain that all of one’s cognitive states can or do influence all of one’s perceptual states. Then they hold up cases like persistent visual illusion as counterexamples to the “continuity” view as characterized, since these are cases where cognitive states apparently fail to influence perception (the Muller-Lyer illusion is a common example). But note the sleight of hand: this would be a counterexample only if the so-called continuity view used ‘all’ instead of ‘some’ to quantify their claims about cognitive influence on perception. They don’t. (To be less cryptic: I think this is precisely how modularists like Fodor and Pylyshyn characterize one of their explicit opponents, Jerome Bruner and New Look psychology. It’s a rather uncharitable characterization, not least because Bruner explicitly disavows such a view in his “On Perceptual Readiness’ (1957), a review piece of decades of New Look research. At least in this dialectical space, it’s remarkable that the Muller-Lyer still garners so much attention.)
J.D : In your book you suggest that the claim that thinking directly influences perceiving is unorthodox in philosophy and cognitive science. Can you briefly say why?
D.S : So for a while now I’ve been an opponent of modular architectures of the mind. It’s important to note that this term, like many, isn’t used in a singular way to identify a unified theory. But the version I resist is the one typically associated with philosophers like Jerry Fodor and psychologists like Zenon Pylyshyn. Even here there isn’t (on my reading anyway) a single unified theory. But the version of the theory that I oppose is the one that says that perceptual systems like vision are “informationally encapsulated”: once the eyes and the rest of the visual system have received their input from the external world, that system processes that information in a way that does not rely upon, and is not influenced by, information elsewhere in the mind. Importantly, this implies that cognitive states like belief, desire, and intention, as well as processes of explicit reasoning, play no role in visual processing. Therefore, in the terms of the modularist, visual experience is “cognitively impenetrable”. I, and several others, have drawn on empirical evidence to argue that this view is mistaken.
I think this view is an orthodox one in philosophy of perception in contexts where one wants to maintain a position like mine. This is evident in the debates about the cognitive (im)penetrability of perception. Opponents of modularity identify some empirical data that they think are best explained as the cognitive penetration of perception, and this purports to be a counterexample to the putative encapsulation of perception. A modularity theorist then rejects the counterexample, offering an explanation compatible with modularity. The process repeats. What I think this betrays is that plausible cases of interesting cognitive influence on perception are always measured up to modularity; their value is determined by whether they suffice as a counterexample to that theory. The underlying assumption is thus that modularity deserves some kind of default theory status. Any warrant for this assumption requires either very strong arguments or strong explanatory power. And I think when one seriously interrogates this version of modularity on both scores, the assumption proves unwarranted. And, finally, since it is a very limiting theory, we have further reason to give it up as any kind of default. (If I’m being honest, I think we should give it up entirely, at least in any broadly scoped version, that is, as an architecture of the mind or of perception. It’s far from clear to me how it is theoretically beneficial. And any observant reader will notice the tension here: if one follows my prescription then one maybe should stop, or not start, reading earlier work by Stokes!)
J.D : A few related questions here: (i) What do you think is the unique contribution of philosophy to the topic of cognitive influences on perception? (ii) Is there an interesting division of labor between philosophy and cognitive science in this respect? (ii) In what ways is your view influenced by current empirical results and hypotheses?
D.S : (i) One straightforward contribution is identified by many philosophers of science: the philosopher can offer conceptual tools, identify and ground important distinctions, reveal problematic assumptions, and the like. I don’t think this is different for the philosopher of cognitive science. And because the emphasis in cognitive science is on mental phenomena, the philosopher can also identify the importance and implications of those phenomena as scientifically theorized. An example of the two combined, in my own work, is this. There are various extant definitions of the cognitive penetrability of perception in relevant literatures which often take the form of the philosopher’s “essential” or “real” definition (I myself have given at least one). These definitions often cross-cut one another, yielding inconsistent verdicts on the same empirical results. I’ve suggested that a way out of this stalemate is to identify why a phenomenon such as cognitive penetrability of perception would matter if it occurred. I then suggest that we characterize (even if not define) the phenomenon accordingly, in terms of consequences. The metric for an interesting case of cognitive influence on perception is whether it implies one or more of the consequences of value to most parties of the debate.
A more recent example from my book is this. If I am right that thinking affects and sometimes improves perception, and that this sometimes comes to intellectual virtues, then we bear a great deal more responsibility for how we perceive the world, and for both the good and the bad cases. This has deep implications for how we understand ourselves and our perceptual contact with the world.
(ii) Yes, I think there is a division of labor, but also healthy overlap. As I intimated above, scientific study is often highly theoretical. And this is even more so for something like psychological modelling. At least at various points, one might even say that such modelling is (partly) philosophical, and this is true whether it’s done by card carrying philosophers or by neuroscientists doing philosophy. Also, I guess I’m a bit old school in thinking that philosophers—at least philosophers like you and me—are cognitive scientists or are at least part of the game. That is, I think of philosophy as included in the nexus of disciplines that makes up the field. Of course few of us are properly trained experimentalists (or programmers or roboticists or linguists), so few philosophers can play those roles but have to defer to others more qualified. But I’ve been lucky enough to collaborate with great people and pick up a small few of those skills up over the years.
(iii) My work is hugely influenced by empirical results and hypotheses, and generally by intentional interdisciplinary engagement. A big part of my project is mental architecture, attempting to determine the right way to categorize mental processes and their relations. So naturally, I rely on existing studies—behavioural and neural in particular—of various mental processes of interest. A big turn in my recent work is the emphasis on perceptual expertise studies. And most recently, I’ve been lucky enough to collaborate with psychologists at my own university—Jeanine Stefanucci, Trafton Drew, and Sarah Creem-Regehr in particular—on studies on varying kinds of perceptual experts. (And years before that, I was lucky enough to collaborate with researchers in artificial life and robotics, most especially Jon Bird). I count myself very, very lucky for these opportunities. It’s one thing to get a working psychologist or computational neuroscientist to talk with you about their work (and do so philosophically), it’s quite another to engage them to develop new experiments with you. (It turns out that experimental design is very difficult, for the record. Who knew!?)
J.D : In your book and articles, you draw several consequences of your claim that the mind is much more malleable than we usually think. One of them concerns art and aesthetic experience. Can you give a few examples of cognitive influences on aesthetic experiences and how this might be relevant (if it is) to how artists (should) conceive of their artworks with respect to both reception and production?
D.S. : Sure, in brief, I think that aesthetic experts are a subset of perceptual experts. In more commonly studied cases of perceptual expertise—say in radiology, sport, or birdwatching—the expert performs exceptionally, reliably making diagnoses with great accuracy, rapidly acting on complex patterns of play, making fine categorical discriminations between stimuli. In some of these cases, this advantage is grounded in improved perceptual skill, where those improvements depend upon what the expert has learned (partly cognitively) in their respective domain of specialization. And what the results of a range of empirical studies suggest, across a variety of domains and task types, is that this performance involves holistic processing: an enhanced sensitivity to patterns, organizational features, gestalts. The robust analogy with face perception here may be helpful: we tend to better perceive faces and their features when they are presented as wholes (and moreover, when they are presented in ecologically valid ways, for example, upright). This also seems true for non-face perceptual expertise.
Now consider a couple examples. The ballet instructor is especially adept at identifying instances of gracefulness or balance in her pupils. And the art historian is especially adept at distinguishing many instances of impressionist painting from paintings in other genres. Now the difference between spotting and not spotting the balanced dance performance or the impressionist painting is plausibly a phenomenological one: recognizing, in experience, an event as balanced is phenomenally distinct from not so recognizing it. Further, it is plausibly a sensory phenomenological difference in one’s visual experience. And it is not (just) that the ballet instructor or impressionist spotter (by contrast to the naïve) perceptually represents distinctive colours or shapes or edges, but instead differently perceives how those basic perceptible properties are organized. She perceives the balance in the performance or the impressionism in the artwork. And this, on my view, is perceptual expertise in the form of enriched perceptual content. Here’s why.
A minimal thesis says that aesthetic properties are nothing more than particular organizations or patterns of properties; (some) aesthetic properties (and artistic properties) are gestalt properties. This extends a view proposed by Frank Sibley, which says that aesthetic properties emerge from combinations of basic physical descriptive properties (Sibley 1959). For substantive aesthetic properties like being serene or dynamic or balanced, there is a distinctive gestalt or cluster of gestalts that characterize such properties. Or, stronger, those organizational or gestalt properties just are the aesthetic properties. Perhaps there is nothing more to being balanced than an organization or gestalt of basic features. And this is what one perceptually recognizes when one experiences something, a dance performance, say, as being balanced. Likewise for artistic properties like 'being impressionist'. This gestalt includes, say, highlighting of natural light and reflection; a regular use of lighter colours; identifiable quick, short strokes of paint, and so on. To know impressionism, is to know and respond to some cluster of these features and how they are organized. It is to enjoy a perceptual sensitivity to the impressionist gestalt.
Notice how well this comports with the general account of perceptual expertise. What the impressionist spotter enjoys is (like the expert radiologist) an enhanced sensitivity to patterns, organizational features, gestalts. And this is a perceptual sensitivity that, at least in some cases, depends upon what the individual has learned in richly conceptual and culturally sensitive ways. And this reveals another important perceptual improvement that can come with acquisition of perceptual expertise: in ways specific to their domains, experts enrich their perceptual content (in that domain). It also suggests, for those interested in the history of philosophical aesthetics, a rather simple way to understand Sibley’s famously elusive concept of taste. He claimed that recognizing aesthetic properties required something more than mere perceptual capacity. In one sense he is right: what it requires is some background cognitive learning (and this can vary by degree). In another sense he is wrong: the expert doesn’t enjoy extra-perceptual capacities in any special or transcendent sense; she enjoys cognitively improved perceptual skills with rich perceptual content. Having taste on my version of Sibley’s account thus comes to having some degree of aesthetic, and thus perceptual, expertise.
The last part of your question is intriguing, but I’m not sure how much I have to say about it yet. If my view is apt, then perhaps artists may want to be aware that some (often a lot) of their audience cannot presently see the work in the same way, since they lack the same visual expertise as the artist. (Descriptively, at some level anyway, this is not implausible given dramatically wide aesthetic judgments and appreciation, across both artists and audiences of art.) I’m not sure how this yields a prescription, but there are certainly two observable extremes. Some artists might be inclined to produce more accessible work (and perhaps to make it more instructive to the audience in some way). And they might do this just in case they want the audience to understand the work in some rich way that tracks their intentions or motives. Alternatively, and for the same reasons, some artists might respond by simply eschewing any such responsibility to the audience. Like them or not, I’m sure that one can think of clear examples at either end of the extreme.
J.D : The notion of perceptual expertise or skill is central to your theory. Are there limits to how thinking can influence perceiving, and to what can be perceptually learned? For instance, when a schizophrenic subject judges that the world is coming to an end by looking at a geometric pattern carpet (an example reported by Chris Frith, if I am not mistaken), has she or he learned to perceive something that we just cannot?
D.S : I’m sure there are limits. For one, the phenomena in question have to be perceivable, at least in the sense that they have to have some visual or other sensible properties. So this makes the view consistent with cognitively influenced illusion and hallucination, but not with cognitively influenced perception of abstractions. However much you may think about it, you can’t visually perceive numbers or sets or universals or kinds or conventions. (This is a good thing for, say, mathematicians and metaphysicians!)
Brief interlude: the latter observation, if correct, has important consequences. If natural kinds are essences or conventions (two distinct, standard views about natural kinds) then they are inappropriate candidates for possible contents of perception. This cuts against the grain of some of the recent work on admissible contents of experience. Instead, and as I noted above, I think aesthetic properties—understood as gestalts, or patterns, or organizational features—are much better candidates for possible perceptual contents, since they can be perceived through the senses.
In my book, I argue that some cases of perceptual expertise manifest epistemic virtue, where that virtue is resident in the experience itself. Experts develop a reliable perceptual skill or disposition, that deploys largely automatically, but is a result of background cognitive etiology. They thus enjoy an intellectual virtue, an improvement to perception, for which they are responsible. With this potential for improvement comes potential for vice (as is standard on a virtue-theoretic account). False beliefs and bad theories can, in principle, change perception for the worse.
One such case that I briefly discuss in the book is implicit bias and the cross-race effect. Most human beings are expert face perceivers: they can recognize and identify other human faces with far greater accuracy than other similarly complex stimuli. With one glaring exception: we are not experts at perceiving human faces from “other” races. Now it is plausible that racial categories are socially constructed; they are parts of a bad social theory. It is also plausible, unfortunately, that because such categories permeate culture that they ground implicit (and explicit) racial biases. Finally, the cross-race effect seems to be both perceptual (partly explained by exposure to cross-race faces) and cognitive (grounded in how we categorize others as, for example, in-group versus out-group). Putting all of these pieces together, the cross-race effect may be a vicious case of culturally sensitive, cognitively influenced visual perception. It is implicit bias (and sometimes explicit, unfortunately) that reaches all the way down to visual perception. One silver lining here is that recent studies have shown that subjects can be trained out of the perceptual effect when cued or primed with information that undermines the biased categorization. The cross-race effect is apparently malleable. (I should note that in a recent book symposium, Jonna Vance has made some very insightful suggestions to me on how we might think about vicious cases of expertise, in particular how we might aim to guard against them individually or collectively. See here).
As for the Frith case, I don’t know if this is a case that relies on cognitive learning in the etiology of the perceptual process, in a way that approximates my cases of expertise. Also I don’t know if it’s a perceptual phenomenon, or rather a difference in the kinds of judgments that the individual makes on the basis of the perceptual experience. But in principle, this kind of case is certainly possible. Again, with the potential for virtue at any level, including the perceptual, comes the risks of vice, or error, or misrepresentation. In saying this, however, I’m not saying that the individual is perceiving something, in any veridical sense, that we cannot. My view is not (I hope!) radically constructivist or relativist in this respect.
**"A note of thanks to Jérôme Dokic for constructing this interview and facilitating my visit (which took two attempts given the pandemic!), and for members of the Institut Jean Nicod and EHESS for having me. I am both flattered and grateful." Dustin Stokes.**