Does skilled perception confer justified beliefs?

This is the fourth installment of my featured author series on skilled perception and other skilled practices. In this post, I want to explore the issue of justification. We frequently rely on specialized perceptual skills, acquired as a result of extensive practice in a given domain. Birders discriminate species of birds on the basis of subtle cues such as size, shape, colors, and habitat. Radiologists can diagnose a patient on the basis of minute differences in grey shade on an x-ray. Primatologists can discern dominance hierarchies and other social relationships where novices only see jumbles of monkeys. Even with the naked eye, jewelers can tell apart gemstones that look identical to neophytes by properties such as hue, saturation and luster. I will refer to such instances as skilled perception.

Some authors, such as Chris Tucker, have argued that skilled perception provides prima facie justification in the same way as ordinary perception does. The example he provides is of an ordinary person, seeing a sea creature, who obtains a justified belief she sees sea creature, whereas an expert, like Jack Sparrow, forms the justified belief that there is a bottlenose dolphin. I want to resist the idea that ordinary perception and skilled perception can lead to justified beliefs in an analogous way.

As I outlined in a previous post, both ordinary and skilled perception feel phenomenologically natural, but their genesis is very different: ordinary perceptual skills come about by normal interactions with our environment, practiced skills require many hours deliberate practice and being taught or apprenticed before they start feeling natural. Do such skills provide us with justification, and if so, how? The peculiar thing about skilled perception is that without the requisite deliberate practice (e.g., hours of observing X-rays during one's medical internship) the relevant perceptual features remain unobservable.

Whether one's perception is truth-tracking is measured by criteria that are inherent to the practice. As Kitcher has argued, this might lead to a worry that skilled perception shows a heavy theory-dependence of observations, making skilled perceivers' interpretations circular. Consider the following example (slightly adapted from Kitcher): Sarah is an experienced behavioral biologist. She observes interactions between members of a baboon troop in a sanctuary. She sees and hears dominance hierarchies, alliance building, aggression, and submissive behavior. She infers who the alpha male is, and what interactions he has with other males and females in the group. Her PhD student, Jason, who is not yet trained in such observations, sees a jumble of monkeys moving in a seemingly random fashion. She points out to Jason what to pay attention to, gait, posture, gestures and grooming, facial expressions.

How can Jason know that he is really learning to social relationships between monkeys? Kitcher argues that experts can engage in “displays of discriminatory virtuosity”: Sarah can predict what will happen next in a social interaction between the chimpanzees, and thus show that her skill really tracks social relationships between the baboon troop. I do not think we need to require that expert skilled practitioners exhibit displays of discriminatory virtuosity. Even in the absence of any information about the skill, its outputs provide prima facie justification. How?

Some philosophers, like Tucker and Pryor, propose that the mere phenomenological seeming of truth (which Huemer calls "forcefulness") that accompanies ordinary perception and intuitions provides a source of prima facie justification. Dogmatism says you can trust the outputs of ordinary perception just because they feel obviously true. The problem is that skilled perception involves a lot more inference than ordinary perception. As Elisa Freschi has detailed, Indian philosophers have argued that a skilled jeweler can identify gems by sight, but infers their value. Similarly, Sarah's observations (in field notes) are a combination of perception and inference, as are the radiologist's diagnosis on the basis of the X-ray. Either we cut out all the non-inferential parts (which would take away essential outputs of the skill), or we need to abandon the dogmatist option.

Recently, Susanna Schellenberg has argued that ordinary perceptual experience provides us with evidence because of its metaphysical structure. Sensory states provide perceptual evidence due to the metaphysical dependence of sensory states in bad cases (hallucinations) on good cases (normal, veridical perception). Simply put, without the outputs in the good case, the outputs in the bad cases wouldn't occur, so there is some prima facie evidence that all perceptual states confer. If I hallucinate that I am seeing a cup, I am doing so because my perceptual system, in good instances, picks out cups.

However, the case of skilled perception is different. In good scenarios of acquired perception, the perceptual outputs are generated by a truth-tracking skillful practice such as birding. But there are many scenarios of skillful practices that don't depend on any relationship with the world: tarot card reading, oracle divination, palmistry, all purport to track some aspect of reality but they don't: this makes all their outputs spurious. Only once you've established the practice has epistemic good-making features, the metaphysical reasoning holds.

Still, while I have shown that metaphysical dependence or dogmatism don't work for skilled perception, the outputs of such practices are still justified in the absence of any information about the practice.

Consider, first, the risk of deception. In human cultural transmission, there is a risk of deception, and perceptual skills are culturally transmitted. The risk is low for skilled epistemic practices since they require an extensive period of apprenticeship. In such cases, honesty has a by-product advantage, and dishonesty is hard to pull off. For example, Sterelny has remarked it is easy and credible for him to present as an Australian birder because he really is an Australian birder. He put in considerable effort to learn the requisite birding skill, but once acquired, he does not even need to signal consciously— the evidence for his birding qualities just arises as a byproduct of his daily life. By contrast, someone who wants to present as a birder without the skills would need to invest lots of time and effort to be able to pass as such, including at least some passing knowledge about birds, and investment in the relevant binoculars and other birding equipment. When I was an archaeology student, one of the optional readings was “Bluff your way in archaeology” written by the renowned archaeologist Paul Bahn. The reason this book was offered was that it actually contained a lot of real information you need to pass as an archeologist, the more accurate title would have been “Gain some elementary knowledge in archaeology so as to pass as an archaeologist”

Second, as we have seen in this post, neuroscientific studies indicate that skilled perception co-opts functionally specialized neural pathways, such as face recognition areas that are deployed to visually discriminate dogs, cars and birds in experts in these fields. There is a continued debate among neuroscientists about whether the face areas are really specialized for faces, or a more generalized functional network that can be trained to do anything, but regardless of the outcome of this debate, it is the case that we use functionally specialized capacities in skilled epistemic practices, which, in a teleofunctional framework also provides reason to think a car expert who sees a 1977 Oldsmobile Cutlass Supreme rather than a 1980 Pontiac Sunbird is really picking up on relevant small differences between these cars (such as the shape of the grilles) to distinguish between them.

Third, in many domains of skilled expertise, the epistemic elements of the skill are part of what it is to perform the skill well. For instance, if an international master looks at a chess position and concludes it is hopeless for white, he is using capacities that are part of what makes a good chess player.

For these reasons, skilled perception yields prima facie justified beliefs, although there can of course be defeaters. For instance, suppose you learn tarot card reading or aura reading, and obtain good external reasons for assuming those aren’t tracking anything they are supposed to track, then you can reasonably come to doubt the validity of those practices and what you have learned through them. (Real domain of such skill might be reading subtle cues in client’s face and posture - successful tarot card readers are good at this.

I'll return next week for a final post on religious experience as skilled perception.

Comments

[Elisa Freschi]

Helen, is (skilled) perception in the case of tarots really (skilled) perception? I would rather say that it is a (chain of) inference(s) which leads the tarot reader from the perception of a given card in a given position to the virdict of, say, "You will meet a tall dark stranger". In this sense, I would rather say that the defeaters regard the possible mistakes in this chain of inferences.

[Michael McNaney]

If the skilled perceptor is realizing their place in the string of causation and if the relatively of time and circumstance of the situation is considered as well as being completely altruistic in their intent then yes, perhaps there is justification.
The real question then becomes are these things part of their skill. Human condition shows all three of these together is nearly impossible.
Creates a bit of a paradox.
Don't you think?

Mac/

[Helen De Cruz]

Hi Elisa: I was thinking of cases where you don't know if the skilled perception is in fact picking out anything (i.e., is a genuine case of perception) or if it is in fact a pseudo-skill that is not tracking anything. While it would seem that tarot card reading is not a perceptual skill because of the chain of inferences, there are lots of skillful practices where the causal connection between what you are picking up and the inferences and perceptions you make are opaque. Take the example of the behavioral biologist, who sees, say, one monkey plucking another monkey's fur and can infer on the basis of that that the one monkey is subdominant to the other. Or the archaeologist who can tell the age of an earthenware pot by looking at the squiggly lines on it. In each of these cases, you accept the chains of inferences you make on the basis of trust (e.g., from experts who tell you that you how to make these inferences). It does not seem clear-cut to me how you can, from the outset, know whether you are really tracking anything (and I've given some reasons for at least placing a prima facie trust in the skills you've learned).

[Elisa Freschi]

Helen, I beg your pardon for insisting (I am just trying to understand). Let us focus on the archaeologist. She sees a given pattern and infers a certain age for the pot. This, I would say, is a plain case of inference. If she tells me that the pot is an instance of black figure pottery and is thus dated to the 6th century, the knowledge I derive from her statement is due to testimony.
The more intricate case, instead, is that of her ability to *see* a given aspect of the pot (say, a hue of colour of the ceramic), which a lay person would not be able to perceive. I would call this perception testimony-enhanced perception in the sense that it is a form of sense perception enhanced by what one knows through testimony (the archaeologist only notices that feature because she already knows about its significance). In your terminology, it is a case of skilled perception. And it seemed to me that you convincingly argued for the reliance on that perception *as if it were* ordinary perception. In the example at hand, this would be the reason why the archaeologist trusts her eyes and it would be ---together with inference--- one of the ultimate sources for the information she gives me.

[Helen De Cruz]

Hi Elisa (sorry for my delay in response!) I find it difficult to separate testimony-enhanced perception and inference in the case of skilled perception since they are so often intertwined (but I have not made up my mind about whether we can draw a principled distinction so I welcome your thoughts on this) In the case of the archaeologist, some of the features have significance because they point to the age of the pot (what you call testimony-enhanced perception), but whether she also "sees" that it is, say, an early linear pottery vessel, or infers it is not so clear to me (thinking about the example of the jeweller you put in your blogpost as well).

[Elisa Freschi]

I would say that "This pot is from the 7th c. bC" is an inference on the basis of some perceptual data (the black figures, etc.). However, I also notice that experts "see" things I do not even notice. My classical example is a stroll in the woods with a botanic expert and a layperson. If asked, at the end, about what they saw, they would answer in two very different ways and the layperson might even react with something like: "Where did you see all these wonderful things? I looked and could not see any of them". In such cases I would say that perception is dependent on the fact of knowing what to look for and how to name it.