Becko Copenhaver: Comments on Stokes

Becko Copenhaver
Washington University, St. Louis

“We are, all of us, potential perceptual experts.”  (Dustin Stokes, Thinking and Perceiving, p. 234)

I’m grateful to discuss Dustin’s Thinking and Perceiving. I agree with him so much, about so much, that I hope our points of departure will be revealing. For example, in the quotation above, Dustin offers us all the promise of perceptual expertise, potentially. My own view is that we are all, already perceptual experts, but about the most mundane things: yellowjackets, stop lights, cupcakes, the quality of air before a storm. Moreover, I hold that much of our perceptual expertise is not cognitive. Finally, on my view, we begin our perceptual education from birth, learning from mothers, siblings, and peers; but our teachers include more than these: we are taught as well by the world itself.

I’m not sure whether or where Dustin disagrees with this picture, so let me set out some points I’m confident we do agree on. We agree that there are important distinctions between:

  • Perceptual development (PD). PD is the maturation of endogenous sensory, perceptual, cognitive processes, that may require exposure to stimuli for their development. PD is genuinely perceptual: developmental changes are changes in experience, including representational, phenomenological, and behavioral changes.
  • Perceptual learning (PL). PL is a long-term change in perception as the result of practice and experience.* The changes need not, though may, be the result of explicit training. The changes include acquiring or increasing sensitivity to differences, unities, weighting, patterns, organization, feature types, and categories. Importantly, PL is a change at the sensory level – it occurs at early stages of perceptual processing. PL is genuinely perceptual: the changes are changes in experience, including representational, phenomenological, and behavioral changes. PL is an outside-in rather than top-down process.
  • Cognitive permeation (CP). CP is a stable short-term or long-term cognitively sensitive change in perception as the result of high-level or cognitive learning or training. The changes in perception include acquiring or increasing sensitivity to differences, unities, weighting, patterns, organization, feature types, and categories. The changes require explanatory appeal to high-level cognitive content acquired by learning or training; they require appeal to belief or knowledge. CP is genuinely perceptual: the changes are changes in experience, including representational, phenomenological, and behavioral changes. CP is a top-down process.

A few clarifications: Dustin refers to CP as ‘perceptual expertise’ by contrast with PL and PD. I would like to leave it open whether some or many cases of PL are perceptual expertise (since I think they are). In addition, by ‘cognitive permeation,’ I mean the same thing as Dustin, and most, do when using the term ‘cognitive penetration’. There’s no difference in meaning; I just find the nomenclature of the rational penetrating the sensory…unsubtle.

Dustin and I agree that all three phenomena occur. We’re both pluralists. I’m no skeptic regarding CP. I agree that “…many cases of perceptual expertise are both richly perceptual and importantly cognitively sensitive (166).” We agree that CP is an intellectual epistemic virtue – it is an “epistemically-enhancing influence on perception.” I hold that PL is also an epistemic virtue, and for reasons I’ll go into below I suspect Dustin does not. We agree that the difference between PL and CP is that the latter is a cognitively sensitive perceptual phenomenon, while the former is not.

What I’m not sure about is how Dustin thinks of the difference between PL and CP aside from the agreed-upon differences. Sometimes it seems that Dustin thinks of PL as “…a strictly passive process of sensory reception…(165)…” where “repeated exposure to a stimulus kind improves one’s ability…to perceive instances of that kind…(109).” On this reading PL is the result of “mere exposure” rather than “content-sensitive training (208).” This puts PL very close to PD with the difference that PD is endogenous.

Other times it seems Dustin is thinking of PL along the lines I find most compelling: that like CP, PL differs from PD in being a:

  • domain-specific acquired perceptual sensitivity to patterns, features, unities, and categories,
  • that shows up in the representational and phenomenal content of experience,
  • that is the result of practice and experience,
  • that gives the perceiver perceptual recognitional capacities to identify novel exemplars,
  • which recognitional capacities do not readily transfer to other, even very similar, perceptual tasks.

This latter picture is another way of being a pluralist about perceptual expertise. Some of our expert experience is, as it were, supply-side: it is supplied at least in part by cognitively sensitive processes. And some of our expert experience is demand-side: it is supplied at least in part by an environmentally sensitive process: a process sensitive to the practical demands of the environment the perceiver (contingently) finds herself in. Like CP, PL requires learning or training– not mere exposure – but the training need not be by instruction (though it may be). The world itself is often our trainer: if there are enough yellowjackets around, the world will train you to be a perceptual expert about yellowjackets.

I prefer this picture on naturalist grounds. Though it may turn out that humans are the only perceptual experts, we ought to begin with the assumption that perceptual processes are more-or-less continuous in nature given the kind of biological function Dustin and I agree perceptual improvement has. I’m not sure that non-human animals have cognitively sensitive changes in perception (though I wouldn’t rule it out). But I’m confident that they have changes in perception as the result of practice and experience that make them differentially sensitive to features and things in their environment that are important in their daily, practical lives.

I wonder whether Dustin would resist this picture. If he does, it might be for epistemic reasons. Dustin argues convincingly that CP is an intellectual epistemic virtue, and he holds that an “epistemic virtue is properly attributed to an agent who is, to some appropriate degree, responsible for the relevant performance (195).” Because PL is most often driven by environmental demands, one might think it could not be an epistemic virtue. It’s the yellowjackets that are responsible for the presence and performance of yellowjacket experts. My sense is that identifying the locus of responsibility is a worry for virtue theory in general rather than a worry specific to Dustin’s project or mine. But my initial response to this general worry is to be a staunch anti-individualist about responsibility. It may be tragic, but the virtues are an ineliminably social, worldly phenomenon. Their acquisition and performance is located in the individual only by abstraction. This is true of CP no less than PL.

Few of us are radiologists or ornithologists or football players, but we’re all perceptual experts: it’s just that our expertise goes unnoticed, even by ourselves. Our mundane expertise allows us to accomplish the routine achievements of living an everyday life.

*My thinking about perceptual learning has been influenced by many people, including Dustin. But I am particularly indebted to Kevin Connolly and Adrienne Prettyman.


  1. Thanks for this comment, Becko, which (as usual) is clarifying.

    I’m still a bit worried about CP. In my comment to Dustin, I wondered whether conceptual teaching changes perceptual experience. My example was one way I could recognize convex shapes—check each point on the outline to determine whether there is a straight line to another such point that goes outside the outline. I could also come to recognize such shapes by learning a non-conceptual “holistically” apprehended feature of convex shapes. I accept that the latter effects genuine perceptual change, since it isn’t the application of pre-existing perceptual skills. But I am sceptical that the “spelled-out” recognition procedure affects perception, precisely because it is an application of perceptual recognition skills I already had. You and Dustin are apparently not sceptical about this.

    There’s another kind of CP that I readily acknowledge. If I have learned through experience that roses are red, I may experience their colour as redder than if it were presented in a different context. This is what one might call doxastic shaping. But I don’t think that it is an instance of Dustin’s TaP or of your CP-perceptual-expertise. To apply your characterization of CP, no “high-level cognitive content” is acquired here.

  2. Becko Copehnhaver

    Thank you Mohan. This allows me to ask you about a view I’m not sure what you’d think about. I have a dictum when thinking about PL and CP:

    Dictum: distinguish learning processes from the changes in perception that are the result of learning processes.

    In the case you describe, two people have the ability to recognize perceptually, convex shapes.

    The first acquired this ability through a learning process that invokes pre-existing perceptual skills (are these learned or not?).

    The second acquired this ability through a learning process that invokes holistic apprehension.

    On your view, the first is not a change in perception — is that right — because it’s just the deployment of pre-existing perceptual skills.

    On my view, whether something counts as a change in perception doesn’t depend on its etiology. I think of PL and CP as developmentally continuous. Learning processes are going to draw from lots of different etiologies from environmental feedback loops (PL) to high-level cognitive processes (CP).

    Say that you’re hired for a job that requires you to be an excellent convex-shape detector. At first, you draw on pre-existing perceptual skills — learned and unlearned. You “check each point on the outline to determine whether there is a straight line to another such point that goes outside the outline.” But eventually you’ll stop doing that. It’s not that you’ll be checking more quickly or covertly. What had been a checking activity gets off-loaded (to use Connolly’s terminology) to the perceptual system. That off-loading is a change in perception.

    Whether a change in perception is due to CP or PL has to do with its origins, of course. Learning processes may involve cognitive permeation, inference, pre-existing skills, judgment, association, habit, etc. But the changes in perception that are the result of these learning processes are not themselves thereby cognitive permeation, inference, pre-existing skills, judgment, association, habit.

    Or have I misunderstood your question? Thank you for your kind words and your contributions to this discussion. I’m learning a lot.

  3. dustinstokes

    Hi both, just a quick reply for now. I agree with Becko’s characterization here (in fact, as will become clear on Friday in my reply, I agree with almost everything in her comments, at least on matters specific to my book). I would subscribe to Becko’s Dictum. And in past work on CP, when I’ve attempted to define the phenomenon, I characterized it as a cognitive-perceptual relation, but the cognitive component and perceptual component are separably determined (assuming some workable cognition/perception distinction). So while cognition may be causally efficacious in bringing out the perceptual change (in fact, this could be a strong counterfactual relation, depending upon the case), what makes the change a genuinely perceptual one isn’t that etiology. That change, if it occurs, is constituted by some change in perceptual representation and/or perceptual phenomenology.

    Shall we meet for drinks and sort the rest out?

Comments are closed.

Back to Top
%d bloggers like this: