Knowledge-how: Future directions
Introduction
The aim of this final chapter will be to outline two important areas for future research related to knowledge-how, in mainstream epistemology and where epistemology intersects with philosophy of mind and cognitive science. The chapter is divided into two sections. Section 8.1 articulates some avenues for future work on the connection between skill in practical knowledge and epistemic internalism, which is roughly the view that what matters for a belief’s justification is entirely factors that are internal to a subject’s psychology. This view contrasts with a view made popular after the Gettier problem according to which a belief’s justification depends on some factors that are external to a subject’s perspective.1 Section 8.2 shows how recent work on extended and group knowledge stands to generate new puzzles about practical knowledge which are suppressed in cases where the traditional intracranial picture of human cognition is taken for granted.
8.1 Knowledge-how, internalism and skill
The internalism and externalism debate in epistemology is situated in a context of rethinking the nature of knowledge. In this section we explain the background to this important debate and then in the next section we apply it to an open question about the nature of knowledge-how. Our goal in this section is to suggest a new avenue of research on practical knowledge that builds upon a productive debate in epistemology of the last several decades.
8.1.1 The Gettier problem and internalism/externalism debate
It is difficult to overstate the significance of the Gettier problem for understanding the development of epistemology over the past fifty years. Prior to Gettier’s 1963 essay ‘Is Justified True Belief Knowledge?’, it was widely assumed that knowledge was justified true belief (JTB).2 According to the JTB model, a subject knows that (e.g.) speckled trout are members of the drum family just in case she believes it, it is true, and she has adequate reason for thinking this. Within the JTB tradition much attention on knowledge was given to the nature of justification. Gettier provided a devastating counterexample to the JTB analysis, which had the effect of turning attention away from the nature of adequate reasons to the nature of knowledge itself.
Here is a Gettier-style counterexample to the claim that knowledge is justified true belief. Suppose you own a fancy new Toyota Prius. You proudly proclaim the joys of Prius ownership and turn up your nose at the troglodytes who own gas-hogging vehicles that get less than 53 mpg. You attend a $150-per-plate dinner sponsored by the League of Justice whose aim is to correct every evil in society. Unbeknownst to you, while innocently partaking in your ritual of superiority, some local hoodlums demolish your new Prius. Inside the banquet hall, sheltered from the chaos outside, the Leaders of Justice announce a surprise winner of a new Toyota Prius. In a turn of fate, at the very moment your Toyota Prius is destroyed, you win a new Toyota Prius.
What should we say about your belief that you own a Toyota Prius? At the beginning of the dinner, you know that you own a Prius. You researched eco-vehicles, went to the dealership, and purchased a new Prius. If you know anything about the world around you, you know that you own a Prius. At the end of the dinner, though, you no longer know you own a Prius. However, your justification hasn’t changed throughout the dinner, your belief hasn’t wavered, and by a stroke of luck, the relevant fact hasn’t changed. What has changed is the way in which your justification hooks up with the relevant fact. At the beginning of the dinner, the reasons you had for believing that you owned a Prius were caused by the fact that you did. After dinner, the reasons are still there but there is not causal route from the fact that you own a Prius to the reasons you have.
The Gettier problem shows that knowledge requires more than justified true belief. It seemed fruitless to many epistemologists to beef up requirements on reasons because the Gettier cases showed that knowledge can come or go when the reasons are the same. Thus, the Gettier problem led epistemologists to reconceive the nature of knowledge. Propositional knowledge should be thought of as a natural relation holding between a belief and the relevant fact, rather than thinking that knowledge was a relation between a belief, a fact, and adequate reason. This led to the development of naturalistic theories of knowledge. For instance, there is the causal account (Goldman 1967; Armstrong 1973), the counterfactual account (Sosa, 1991), the truth-tracking account (Nozick 1981), a reliability account (Goldman 1979), or a proper-functionalist account (Plantinga 1993a).3
One main objection to such naturalistic views is that whether or not one knows is not reflectively accessible. If knowledge implies the right to be confident then, on many post-Gettier accounts of knowledge, one cannot tell by reflection alone whether one has this right. For example, on a simple causal account of knowledge, one knows that there is a chunk of gold ore on the table if the fact causes one to have the relevant belief. But there are alternative causal paths to the same belief, namely, when one sees iron pyrite. Consequently, because some causal paths to the same belief are ‘good’ and some are ‘bad’, one cannot determine by reflection whether or not one has the right to be confident that there is gold ore before one. The correct response seems to be to deny one has knowledge rather than thinking that in the good case one has a right to be sure and in the bad case one does not.4
An early exposition of a causal account is by Armstrong (1973). Armstrong describes this kind of account as the ‘thermometer model’ of knowledge. This model compares non-inferential knowledge-that p (knowledge not based on reasons) with the outputs of a reliable thermometer. When a thermometer indicates an increase in temperature by displaying a change from a reading of ‘76’ to a reading of ‘80’, this is (normally) caused by an increase in temperature from 76 to 80. The thermometer analogy suggests that non-inferential knowledge is belief that is caused by the relevant fact in one’s environment. Armstrong gives us a new post-Gettier conception of knowledge. Knowledge isn’t to be understood in terms of beliefs based on adequate reasons a subject possesses but rather in terms of beliefs caused by the appropriate fact.
It didn’t take long for epistemologists to realize that there was something wrong with Armstrong’s thermometer model of knowledge. BonJour (1980) describes these new post-Gettier accoun ts of knowledge as ‘externalist’ accounts. He provides the following counterexample to them, known as the case of Norman, the Clairvoyant.
Norman, under certain conditions that usually obtain, is a completely reliable clairvoyant with respect to certain kinds of subject matter. He possesses no evidence or reasons of any kind for or against the general possibility of such a cognitive power, or for or against the thesis that he possesses it. One day Norman comes to believe that the President is in New York City, though he has no evidence either for or against his belief. In fact the belief is true and results from his clairvoyant power, under circumstances in which it is completely reliable.5
Norman’s belief that the President is in New York is caused by the appropriate fact, but it acts through a power that Norman is unaware of. This power of clairvoyance produces in Norman a belief unsupported by any reasons within Norman’s ken. There is a widespread judgement that Norman does not know that the President is in New York.
BonJour’s Norman case motivates the idea that knowledge requires reasons. As a consequence, the thermometer model of knowledge is inadequate. In the 1980s and 1990s the upshot of the Norman case and others like it (e.g. TrueTemp, New Evil Demon, Chicken Sexer) were vigorously contested. This debate is known as the ‘internalism/externalism’ debate in epistemology. Internalists defend the claim that knowledge requires reasons, where reasons are normally understood as experiences or beliefs within a subject’s perspective that bear on the truth of the relevant claim. Externalists, by contrast, deny this claim, usually arguing that internalism implies scepticism and that internalism doesn’t fit with normal attributions of knowledge to small children. Internalists reply, but tracking this debate would take us to far afield from our current goals.6
8.1.2 Internalism and skilful action
The internalism/externalism debate in epistemology focuses on whether propositional knowledge requires reasons. We propose to offer a rapprochement between internalist and externalist views by focusing on knowledge-how rather than propositional knowledge. When one knows how to ride a bike one has the skill of bike riding. Riding a bike is not something that happens to a person; it is something the person does. Moreover, when riding, a person knows what she is doing and why she is doing it. Her action is integrated within her perspective.
The point here is not that a competent bike rider is always attending to what she is doing; rather the point is that she exhibits a unique kind of awareness of her actions. Let us call this kind of awareness ‘recognitional-awareness’. We distinguish recognitional-awareness from noticing-awareness. In the bike case, one may notice applying greater pressure to the pedals when thinking explicitly about how hard it is to get up this hill and what needs to happen to get up the hill. But the crucial point is that competent action is not normally like that. A competent bike rider takes into account the changes of terrain with appropriate action without explicitly noticing what one is doing. Around a tight curve, a good rider leans into the curve and lays off the pedals. The rider recognizes her environment and responds appropriately. The rider intelligently moves her body in response to her environment. This is recognitional-awareness. One knows what one is doing and why one is doing it, even though one isn’t explicitly attending to the specific movements.
As was noted in Chapter 2, a lesson that virtue reliabilists such as Greco (e.g., 2010) gleaned from the meta-incoherence cases is that knowledge must arise from stable traits that are (unlike Norman’s) cognitively integrated. This point can be understood in a way that complements the moral from such cases that has been drawn by the internalist. After all, one very natural explanation for why Norman’s clairvoyant process fails to be appropriately integrated into his cognitive character is that he lacks any conception of the process that generates his belief that the President is in New York. Consequently, Norman lacks any reflectively accessible reasons for why his belief is correct.
Let us call this feature of knowledge-how ‘its internal character’. The internal character of knowledge-how distinguishes it from ‘knacks’. A golf novice – say, someone who has played only once before – may discover she has a knack for hitting a golf ball out of a sand trap. Such a knack is an ability that is not integrated within the subject’s perspective. The ‘knack golfer’ does not know how she manages to hit the ball out of the sand trap, even though she notices that she can do this fairly reliably. It’s not a skill; it is just a knack.
A person with a knack does not know what she is doing to achieve an end. The golf novice does not know how her actions contribute to the end being achieved. Knacks are different from dumb luck because a knack is reliable whereas dumb luck isn’t. A person who by dumb luck hits a ball out of a sand trap into the hole will be unable to repeat this feat with any success. So knacks are distinguished on one end by skills and on the other end by dumb luck.
Let’s now consider a genuine case of knowledge-how. Mikhail Nikolayevich Baryshnikov, an expert Russian dancer, knows how to perform the traditional Tropak. Consider his performance. He knows what he is doing and why he is doing it. Baryshnikov is not blindly following some pattern; rather his movements are guided by his thoughts. Although he is not noticing his specific movements, if asked to explain why he moved his arm thus he is ready with the explanation. Baryshnikov has recognitional awareness of his movements and his environment.
Contrast Baryshnikov’s performance of the Tropak with Norman the Clairyovant. Norman believes that the President is in New York but he does not know why he believes this. Norman lacks crucial self-knowledge. Another difference lays in agency. Baryshnikov’s dance is an expression of his agency. He aims to perform the Tropak and successfully carries it off. Norman’s belief that the President is in New York is not an expression of Norman’s agency; it is something that happens to him. Norman’s clairvoyant ability is more akin to a knack than a genuine skill.
Baryshnikov’s performance is mindful. This feature of his performance is not a case of his actively reflecting on some regulatory proposition, namely, ‘first, this; second, that; etc.’ Recall in Chapter 1 our discussion of the first Rylean regress. There we observed Carl Ginet’s oft-cited example of an unreflective exercise of propositional knowledge. Ginet writes:
I exercise (or manifest) my knowledge that one can get the door open by turning the knob and pushing it (as well as my knowledge that there is a door there) by performing that operation quite automatically as I leave the room; and I may do this, of course, without formulating (in my mind or out loud) that proposition or any other relevant proposition.7
Ginet’s point is that an exercise of propositional knowledge does not require that one thinks to oneself ‘this is a way to do this’ and then executes those instructions. In our terminology, Ginet’s point is that an exercise of knowledge does not require a noticing-awareness. But when Ginet says that the action is automatic, it is clear that he does not mean that the action is blind. It’s not a mystery why Ginet got to the door and turned the knob; he did that because he knows how to open a door and knows when to do it. The act of turning the knob is mindful.
Both in Ginet’s case and in the case of the Russian dancer, the actions are mindful. It is not a mystery to the Ginet what he is doing by turning the doorknob and it’s not a mystery to Baryshnikov why he moves the way he does in performing the dance. By contrast, it is a mystery to Norman the Clairvoyant why he believes the President is in New York.
We suggest that the difference between reliable abilities that are mindful and those that aren’t is the difference between knowledge-how and knacks. The distinction between knowledge-how and knacks is a new way to frame an internalist/externalist debate with respect to knowledge-how. It is plausible that knowledge-how requires mindful action, action that is integrated within a subject’s perspective. It is not, however, completely implausible that internalism about knowledge-how is false. Consider the following case.
A talented neuroscientist discovers an area of the brain that suitably stimulated mimics the activation of someone who knows how to walk across a tightrope. The neurosurgeon discovers that patients who receive multiple stimulation to this area of the brain retain this pattern of activity for future deployment. Experiments confirm that such subject acquire the ability to walk across the tightrope even though they report not knowing how they do this.
This is a case of reliable action that is dissociated from one’s perspective. The subject does not know what she is doing, nor why she is doing it. Our view is a case is more akin to standard cases of knacks. Once this knack integrated within a person’s perspective, it constitutes a genuine skill. The core issue here is whether knowledge-how requires self-knowledge.
The issue over internalism and externalism with respect to knowledge-how is over whether the following is true:
INT know how A subject knows how to φ only if the subject knows what she is doing in φ-ing and why she is doing it.
The cases we’ve considered provide some evidence for INT know how. Another area of support comes from our discussion in Chapter 6 on linguistic competence. Recall the case of the Turkish tour guide who spoke in perfectly clear English, without understanding a word of English. The tour guide lacked a conception of how the individual words she spoke contributed to the goal of conveying the information she in fact conveyed. To understand a language requires that one knows what one is doing by using specific words and understanding why one is doing it. That is, one understands how the specific words contribute to the communicative goal. Thus we see here too with linguistic competence that there is evidence for internalism about knowledge-how. What we have said here, though, is just the beginning of what is a fruitful area of research.
8.2 Knowledge-how and active externalism
According to externalism about mental content, what your beliefs and other intentional mental states are about – water, chairs, arthritis, trousers, etc. – isn’t just a matter of what’s going on in your head. It is at least partly a matter of how things stand outside your head, for example, in your physical and/or social-linguistic environment. More carefully, the position is that, for a given subject S, the contents of S’s mental states fail to supervene upon S’s intrinsic physical properties.8 And what this means is that your internal physical duplicate, exactly like you in all physical respects, needn’t share your beliefs.
Content externalism is now a popular position in contemporary analytic philosophy, thanks largely to influential thought experiments – such as the twin-earth experiment (see Chapter 4) – in the 1970s and 1980s, due to (among others) Saul Kripke (1980), Hilary Putnam (1975) and Tyler Burge (1986).9
Much more controversial by contrast is a related form of externalism, proposed in the late 1990s by Andy Clark and David Chalmers (1998).10 On Clark and Chalmers’ brand of externalism – what they call active externalism to distinguish it from the comparatively more ‘passive’ content externalist thesis – parts of the external world can do much more than merely individuate mental contents. Parts of the world, for example, a notebook or an iPhone, can under certain conditions partly constitute cognition by functioning as material realizers of cognitive processes.
In this section, we will engage with two distinct forms of active externalism that have been gaining traction, especially over the past decade: the extended cognition and distributed cognition theses. Paradigmatic examples of the former are so-called extended memory cases, such as Clark and Chalmers’ classic example of an Alzherimer’s patient who, in order to compensate for failing memory, relies on a notebook for information storage and retrieval. Paradigmatic examples of distributed cognition include scientific research teams and transactive memory systems.11 In each kind of case, the supervenience base of the relevant cognitive process at issue is not exclusively any individual biological organism.
The specific question we explore is whether extended and distributed cognitive systems attain know-how, and if so, under what conditions? Are there new epistemological issues that arise in accounting for extended and distributed knowledge-how, and if so, what are they? And how would such problems differ from problems framed at the individual level? In the course of developing some tentative answers to these questions, we raise some new puzzles which hopefully future work on knowledge-how will be able to answer.
8.2.1 Knowledge-how and extended cognition
The bounds of cognition
In order to appreciate just what the proponent of extended cognition is claiming, and why, it will be helpful to start with a simple and uncontroversial case of cognition.
inga: Inga relies on her biological memory in a way that is perfectly normal; new information she learns is stored in her biological memory, which is also what she consults when she needs some old information.
Of course, biological memory is intracranial: stored within the skull and skin. Inga’s cognitive process that includes storing and retrieving information takes place entirely within her head. It’s tempting to generalize here about cognition as such: perhaps, as Fred Adams and Kenneth Aizawa (2009) argue, cognitive processes categorically take place in the head, specifically, cognitive processes are brain-bound.12 Call this thesis, which is consonant with folk psychological judgements about the boundaries of the mind, cognitive intracranialism.
Being ‘brain bound’ usually means taking place inside the skull, because brains are usually – in fact, almost always – located there.13 But a moment’s reflection shows that our intuitive judgements about what counts as a cognitive process do not really depend on this. After all, suppose Inga’s brain were enclosed in a portable brain-in-a-vat that she carries around with her, and that it plays the very same role for her regarding information storage and retrieval as it did before.14 In such a case, the process of information storage and retrieval that plays out in Inga’s recently relocated exobrain is surely a cognitive process no less than it was prior to her brain’s relocation.
To the extent that this is right, the key insight of cognitive intracranialism – if it is to be charitably interpreted – perhaps should not be framed in terms of location, but rather, material constitution. On this view, the ‘cognitive processes take place in the head’ slogan should be read as the claim that the material realizers of cognitive processes exclude anything external to the biological brain.
Here we should briefly note that the cognitive intracranialist cannot save the relevance of location to cognition by restricting the location claim to the claim that cognitive process must take place inside the brain. While this revision gets the right result in the revised (i.e. portable brain-in-a-vat) version of the Inga case, it’s nonetheless too inclusive as a general thesis about cognition. Consider that the process of calculating the square root of 2 to 100 digits, when executed entirely by my computer, is not a cognitive process. A Raspberry Pi Zero computer, about the size of a stick of gum, could be implanted in your brain and execute this task within seconds. In this case, the process of calculating the square root of 2 to 100 digits would take place inside your brain, but it would not be a cognitive process.
That said, it’s hardly clear that the revised idea framed in terms of material constitution can withstand objection. As Carter and Kallestrup (2016) have noted, we can imagine that part of Inga’s biological memory – namely, her dorsal premotor cortex – is replaced in the very same location with a silicon-chip device which plays the same functional role in terms of storing and retrieving information. In such a case, it seems counterintuitive to insist that the artificial implant, despite its functional role, is not part of the cognitive process that Inga uses in remembering events that occurred before the surgery. But if that’s right, then it’s not evident that either physical location or material constitution is really crucial to something’s being a part of a cognitive process. Consider now a ‘combined case’, namely, Clark and Chalmers’ (1998) famous case of Otto:
otto: Otto suffers from Alzheimer’s disease, and like many Alzheimer’s patients, he relies on information in the environment to help structure his life. Otto carries a notebook around with him everywhere he goes. When he learns new information, he writes it down. When he needs some old information, he looks it up. For Otto, his notebook plays the role usually played by a biological memory.
It seems obvious here that Otto is using his notebook in a way that, with respect to information storage and retrieval, is on a kind of functional par with the way ordinary agents rely on a working biological memory. Is there a reason we should not count his notebook as part of his memory process? It looks as though to exclude Otto’s notebook from his memory in a principled way, we’d have to (despite the issues raised in the previous examples) somehow insist that either location or material constitution really does matter, so as to disqualify the notebook.
One natural line of critique at this juncture proceeds as follows: an individual’s proper memory processes cannot be duplicated. But Otto’s notebook can be duplicated, and duplicated over and over again. There seems to be little reason to pick one of the many copies as genuinely part of Otto’s memory process. A rejoinder on behalf of the proponent of extended cognition will insist that, to the extent that one is tempted to reject the claim that memory processes can’t be duplicated, this is an artefact of a prior commitment to cognitive intracranialism. More carefully: it’s an accidental property of biological memory that it is induplicable. However, lest we are already wedded to the view that only biological memory processes are memory processes, we needn’t accept that memory processes are not duplicable.
At any rate, as Clark and Chalmers see it, doubling down on intracranialism in this sort of way would reveal an unprincipled kind of ‘bioprejudice’. Their view is that our theorizing about what sorts of things can feature in cognitive processes should be guided by what they call the parity principle:
Parity Principle: If, as we confront some task, a part of the world functions as a process which, were it to go on in the head, we would have no hesitation in accepting as part of the cognitive process, then that part of the world is part of the cognitive process.15
With reference to the parity principle, since Otto’s notebook plays the same functional role as biological memory plays for Inga, we should count Otto’s process of consulting his notebook as a part of a cognitive process so long as we count Inga’s process of c onsulting her biological memory as part of a cognitive process. And if this is right, then Otto’s memory process is one that, as Clark puts it, criss-crosses the boundaries of Otto’s brain and the world. His notebook is, literally, a part of his mind.
Extended knowledge-how?
A defence of the extended cognition thesis is beyond what we can do here. What we want to examine now is how it might matter for debates about knowledge-how, if the view were to be taken on board as it increasingly has been in the philosophy of mind and cognitive science. To this end, it will be helpful now to distinguish between the extended cognition thesis and the closely related extended mind thesis, which are unfortunately often run together. One source for the propensity to conflate the two is that Clark’s original model of the extended mind thesis maintained that beliefs are dynamic rather than static, which is contrary to the orthodox view. If beliefs are viewed as essentially dynamic, then the difference between the extended mind and extended cognition thesis is not an important one to emphasize. However, on traditional thinking, where beliefs and other mental states are not regarded as processes, the difference matters. Extended cognition proponents are not committed, against the background of the traditional thinking about beliefs, to regarding beliefs themselves as supervening partly on things like notebooks and iPhones.16
The difference between the extended cognitive thesis and the extended mind thesis can be captured in terms of supervenience. Supervenience (construed generically) can be understood as follows: A properties supervene on B properties if and only if something cannot change in its A properties without also changing in its B properties. Consider, for example, a pointillism as a painting technique (e.g. Seurat’s A Sunday Afternoon on the Island of La Grande Jatte). The scene that Seurat conveys depends on the individual dots of paint such that one can’t change any of the properties of the scene without changing the individual dots. But, crucially, the properties of the scene – that a dog appears in the painting – isn’t reducible to the individual dots. There are a number of different ways to paint a dog into the picture.
Supervenience is an important philosophical relation. It is used to capture the relationship between moral properties and natural properties. For instance, the claim that moral properties supervene on natural properties commits one to the view that there can be no moral difference between two situations without those situations having at least some natural difference. The advantage of this view is that moral properties are not reducible to natural properties and so they are different but yet moral properties clearly depend on natural properties.
In the case of extended cognition and the extended mind theses, the difference stated in terms of supervenience is the following: whereas the extended cognition thesis says that cognitive processes can supervene partly on extra-organismic parts of the world (such as Otto’s notebook), the extended mind thesis insists that mental states such as beliefs can supervene on extra-organismic parts of the world. This difference is reflected in two different things we could say about Otto’s notebook. According to the extended cognition thesis, Otto’s memory process – namely, the process of storing and retrieving information – supervenes partly on Otto’s notebook; it is a process that does not play out entirely within the brain. This does not entail the further thesis, embraced by proponents of the extended mind thesis, that Otto’s beliefs supervene on the notebook. For the proponent of the extended mind thesis, just as we attribute to Inga dispositional beliefs in virtue of information stored in her biological memory, so we should likewise attribute to Otto dispositional beliefs in virtue of information stored in his notebook.
With that distinction in mind, we’re now in a position to ask the following question: What would it take for knowledge-how to be ‘extended’? How to answer this of course turns on what we mean by knowledge-how being extended. Fortunately, there’s a straightforward way to think about this. Just as the extended cognition and extended mind theses can be captured and contrasted in terms of supervenience, we can likewise frame the notion of ‘extended knowledge-how’ as a supervenience claim. A state of knowledge-how is extended just in case it supervenes at least partly on some extra-agential part of the world. The term ‘extra-agential’ here can refer not just to such things as notebooks and gadgets, but also in principle to other individuals apart from the agent in question.17
Of course, what counts as the supervenience base for knowledge-how differs depending on whether you ask an intellectualist or an anti-intellectualist. And so therefore what extended knowledge-how would involve on each model will be different. There is a further relevant distinction due to Bengson and Moffett (2011a) which was discussed in Chapter 1, regarding (i) being in a state of knowledge-how; and (ii) exercising knowledge-how. Given that the state of affairs of (i) needn’t involve the state of affairs of (ii), the supervenience base of exercising knowledge-how is not identical to the supervenience base of being in a state of knowledge-how. And this means that (for the purpose of whether knowledge-how could be extended in the sense that it could at least partly on some extra-organismic part of the world) it matters whether what’s at issue is being in a state of, or exercising, knowledge-how. In this chapter, we’ll focus on the state angle of the question, as it is most straightforward.
On the reductive intellectualist model, Stanley (2011a) and (2011b), knowledge-how supervenes on propositional attitudes – namely, there cannot be a difference with respect to an agent’s knowledge-how without a difference in her propositional attitudes. Given that the extended cognition thesis does not entail that propositional attitudes (or more generally mental states) supervene partly on items external to the organism, it follows that reductive intellectualism, combined with the extended cognition thesis, do not jointly suffice to yield extended knowledge-how. What reductive intellectualism must be paired with in order to get extended knowledge-how is, rather, the extended mind thesis. If the extended mind thesis were true, then we could envision propositions the knowledge of which suffices for intellectualist knowledge-how supervening partly on things such as notebooks and iPhones.
As several commentators have noted, the extended mind thesis faces several serious objections which are not faced by the extended cognition thesis. One notable such objection concerns ‘cognitive bloat’. If (in short) Otto’s beliefs are in the notebook, then why not – as Sean Allen-H ermanson (2013, 792) puts it – in the yellow pages or the internet?18 Proponents of the extended cognition thesis, by contrast, needn’t engage with this worry about an explosion of beliefs. Moreover, as Orestis Palermos has argued in a series of papers (2011, 2014a, 2014b), extended cognition, though not the extended mind thesis, can be supported entirely on the basis of dynamical systems theory, and needn’t involve (as the extended mind thesis does) any appeal to functionalism.
These considerations suggest that if knowledge-how could be extended, the most viable route would be through the extended cognition thesis, rather than the extended mind thesis. Interestingly, whereas intellectualism, as we’ve seen, requires the extended mind thesis in order to generate extended knowledge-how, things are different for anti-intellectualism. The anti-intellectualist, at least in its most straightforward form, embraces dispositionalism: the thesis that being in a state of knowing how to do something supervenes on dispositions or abilities of the agent; there is no change in one’s knowledge-how state without a change in her dispositional states. So a state of knowledge-how is extended, on a dispositionalist anti-intellectualist model, provided that some dispositions supervene at least partly extra-organismically.
What does this involve? Here some delicacy is needed. When you are in a state of knowing how to calculate the tip when the cheque at dinner arrives, you of course need the cheque; you cannot calculate the tip if the waiter or waitress does not give it to you. And the cheque is something outside your head. Does this mean that, for the anti-intellectualist, your knowledge-how supervenes partly on the cheque, simply because your being disposed to calculate the cheque correctly requires that you have access to the cheque itself? In short, no.
Here it will be helpful to follow Ernest Sosa’s (2015) distinction between the seat, shape and situation (or, ‘triple S’ structure) of a competence. Take, for example, the competence to play Beethoven’s Moonlight Sonata on the piano. If competence consists in being able to perform the Moonlight Sonata on a piano on a certain occasion, then the seat is the basic piano skill for hitting the right notes, the shape is the condition of the individual (i.e. awake, sober, etc.), and the situation is being at a working piano, in conditions suitable for its operation (i.e. not underwater).19 The complete competence requires not just the seat, but also the shape and situation. The situation obviously will involve some extra-organismic things. In this case, it involves a piano.
Most dispositions that are abilities will trivially count as supervening on extra-organismic parts of the world if the relevant supervenience base for the disposition is taken to include seat, shape and situation. What’s of philosophical interest, in connection with the possibility of extended anti-intellectualist knowledge-how, cannot be whether the triple-S structure of a complete competence supervenes partly on items external to the organism. After all, no form of active externalism is required in order to get this result!
The more interesting question is whether the seat, or as Sosa calls it ‘the innermost competence’ – supervenes at least partly on items external to the organism. Notice that, even if you take away the physical piano from a composer, the composer will retain the seat of her competence. The composer will remain such that if a piano was present then (provided the composer is awake and alert) she can perform the piece. In a case where the innermost competence (and not merely the complete competence) supervenes at least partly something external to the organism, the removal of something external to the organism would rob the individual not just of the complete competence to perform an action on an occasion, but also their innermost competence.
The hypothesis of extended cognition, paired with anti-intellectualism, can generate this result. Let’s consider now an analogy, again using Inga and Otto from our previous examples. Suppose that Inga, who stores and retrieves information entirely via biological memory, knows how to use an ATM machine to withdraw money from her bank account. The standard anti-intellectualist insists that this will be a matter of some ability that Inga has. Availing ourselves to Sosa’s distinction, the seat of Inga’s ability to withdraw money from an ATM is her disposition to (when in proper shape, at a working ATM machine) go through the right movements so as to withdraw money. If you take away the ATM machine, Inga still has this innermost disposition, even if she lacks the complete competence to withdraw money on that occasion. However, importantly, if a mad scientist removes from Inga the memory she has of her PIN number, then she loses even this innermost competence. For then, even if she were alert, awake, sober and taken to an ATM machine, she would be unable to withdraw money.
Now, to use an example from Carter and Czarnecki (2016), suppose Clark and Chalmers’ hero Otto is just like Inga in all respects related to ATM machine operation, except for just one difference: that he stores his PIN number in his notebook, rather than as Inga does in biological memory. As an initial point, the anti-intellectualist who embraces extended cognition will want to insist that Otto knows how to withdraw money from an ATM machine no less than Inga does. After all, by the lights of extended cognition, this difference between the two cases is an entirely irrelevant one, in that both ways of storing their PIN number are, as forms of memory, equally valid. Thus, the anti-intellectualist who embraces extended cognition should be prepared to regard the seat of Otto’s memory ability, his innermost competence, as one that includes the notebook, given that the seat of Inga’s memory ability, her innermost competence, includes her biological memory. And if the foregoing is correct, then (for the anti-intellectualist who embraces extended cognition), what results is that Otto’s knowledge-how to withdraw money from the ATM supervenes at least partly on his notebook, no less than Inga’s knowledge-how to withdraw money from the ATM supervenes at least partly on her biological memory.
Taking a step back, the upshot of this result is that anti-intellectualist model, in comparison with intellectualist model, offers a less controversial way to account for the possibility of extended knowledge-how in a philosophically interesting sense. And this is because anti-intellectualism requires merely the extended cognition thesis, but not the comparatively more controversial extended mind thesis that was shown to be needed for the intellectualist to generate this result.20 For philosophers sympathetic to active externalism more generally, this is a prima facie point in favour of anti-intellectualism. However, even for those sympa thetic to extended knowledge-how, as construed along anti-intellectualist lines, the possibility of extended knowledge-how invites at least two new problems.
The Extended Luck Problem In Chapter 3, it was suggested that an important consideration in favour of anti-intellectualism was that knowledge-how seemed to be compatible with two varieties of epistemic luck which are widely taken to be incompatible with all forms of propositional knowledge. These were intervening epistemic luck (e.g. as in standard Gettier cases) and environmental epistemic luck (e.g. as in barn facade cases). However, the kind of extended knowledge-how that results from a pairing of anti-intellectualism and the extended cognition thesis threatens to complicate this picture. For example, suppose that we run a twist on the case of Otto and the ATM machine just considered. Just suppose a jokester tampers with all of the entries in Otto’s notebook, but just so happens to overlook the PIN number that Otto needs to withdraw money from the ATM21 . Does Otto know how to withdraw money from the ATM? The intuition here is a bit murky. For one thing, it’s not as simple as diagnosing this as a case of environmental luck. For while the notebook is in one sense in Otto’s ‘environment’, it is also, in another sense, internal to him – it is part of the seat of his ability to withdraw money. Anti-intellectualists who claim a luck-based advantage over their intellectualist adversaries will at minimum need to develop a principled way of thinking about such cases, and it’s not evident that there is a straightforward way to do this.
A New Value Problem The prospect of extended knowledge-how – in the sense shown to emerge from a pairing of standard anti-intellectualism and the extended cognition thesis – generates a perplexing problem about epistemic value. The problem can be stated simply. In the contrast case of Inga versus Otto, it seems, at least prima facie, that Inga is better off than Otto is, that Inga’s know-how is in some way preferable to Otto’s extended knowledge-how, even though the exercise of both generate the desired result (using the ATM effectively) just the same. However, the pairing of anti-intellectualism and extended cognition, in so far as it generates the possibility of extended knowledge-how, appears to lack in principle the resources for accounting for why we should ever prefer to be in Inga’s situation than Otto’s. The dilemma for the proponent of such extended knowledge-how will accordingly be to either (i) vindicate this asymmetrical value intuition and then explain how such a vindication can be reconciled with the parity principle; or (ii) explain why this intuition is misguided, despite initial appearances.
8.2.2 Collective epistemology, distributed cognition and group know-how
Our ordinary practices of knowledge attribution include attributions that take the form ‘X knows p’ where ‘X’ is a group. Here are three typical examples. (1) Prior to 2015, CERN didn’t know the Higgs Boson existed, but they do now. (2) The jury that deliberated after Ted Bundy’s murder trial knew he was guilty. (3) We want to know what the FBI knew prior to 911.
One kind of response to such talk is fictionalism: the view that it’s convenient to talk as though groups can know things, but that strictly speaking, all such attributions are false because only individuals can know things. Another related response is summativism, according to which, as Anthony Quinton (1976, 19) puts it, to ‘ascribe mental predicates to a group is always an indirect way of ascribing such predicates to its members’. On the summativist view, claims of the form ‘X knows p’ when true, are true just when most of the individual members of X believe that p.
There is, however, a much more interesting way to think about certain kinds of group knowledge attributions. As Jennifer Lackey (2014b, 282) has remarked, ‘A fairly common view in current work in collective epistemology is that groups can have knowledge that not a single one of its members possesses.’22
Consider for example a much-discussed case, due to Edwin Hutchins (1995), of a ship crew navigating a ship to port. A condensed presentation of the case, due to Lackey (2014b, 282), is as follows:
The ship’s behaviour as it safely travels into the port is clearly well-informed and deliberate, leading to the conclusion that there is collective knowledge present. More precisely, it is said that the crew as a whole knows, for instance, that they are travelling north at 80 miles per hour, or that the ship itself knows this, even though no single crew member does.
The knowledge here is, put crudely, a function of the individuals each doing their part, where the justification possessed by each individual is justification regarding some contributory aspect of what it takes for the crew to know they are traveling at 80 miles per hour. What it is to ‘do their part’ is debatable among those who embrace non-summativist group knowledge.23
Presumably, if a group knows some proposition then a group could also know how to do something, even if no individual in the group knows how to do that thing. Indeed, as we noted in Section 5.3.2, Alexander Bird considers it common to indicate that a group knows how to do something. He provides this case: ‘North Korea knows how to make an atomic bomb.’24 Palermos and Tollefsen(ming) agree, recently arguing that there are cases where it seems a group clearly knows how to do something that would be implausible to suppose any individual could do alone. They offer the following case:
corvette: Each individual in the company knows their own domain but no one person knows how to do all the various things that comprise making the Corvette. And this is so even if one reduces individual know-how to individual propositional knowledge. Each individual knows of a way, W, that W is the way to do X where X is her job. But no one individual, we can imagine, knows of a way, W that is the way to make a Corvette because no one individual has all the relevant expertise required in order to build a Corvette. Corvettes are made but apparently no one knows how to make them. This is counterintuitive and it certainly clashes with our practice of praising and blaming Corvette for its cars. The company routinely wins awards for its cars. The credit is given to the company. But on [sic. the summativist] approach no one should be given credit because no one knows how to make a Corvette.
The thrust of Palermos and Tollefson’s line of reasoning can be put as follows:
1 How to make a Corvette is not unknown.
2 If summativism is true, then how to make a Corvette is unknown.
3 Therefore, summativism is false.
The best way to account for the fact that it’s not unknown how to make a Corvette, they suggest, is to credit the company – Corvette – with this know-how, know-how which cannot be attributed to any individual employee. But if Corvette knows how to make a Corvette, what exactly grounds this know-how?
Here, some interesting questions for future research emerge. For one thing, should whatever view is embraced about know-how at the individual level be expected to apply to know-how at the collective level? For example, should intellectualists such as Stanley and Williamson be committed (in virtue of their individualistic intellectualism) to the view that if Corvette knows how to make its signature car, then this is because the group collectively knows some proposition? Or, alternatively, should this kind of ‘univocal’ constraint be relaxed. That is, might it be that (for example) intellectualism about know-how at the individual level could be unproblematically paired with anti-intellectualism at the group level (or, vice versa)?
Secondly, even though there is already some precedent in collective epistemology for what would be required for a group to collectively believe a proposition,25 the nature and characteristics of collective abilities (of the sort the anti-intellectualist is interested in) is less well studied. Anti-intellectualist proponents of non-summativist group know-how must be able to provide a rigorous account of the kinds of realizers of the group abilities that stand to ground collective know-how. As we saw in Chapter 2, and also in Chapter 4, the kinds of abilities germane to knowledge-how must be cognitively integrated, namely, in a way that mere fleeting processes are not. Recall again Bengson and Moffett’s case of Irina from Chapter 2; even though Irina was able to perform a salchow due to a neurological abnormality that led her to perform the correct sequence of moves despite believing she was performing different moves, she failed to know how to perform a salchow. The explanation, it was suggested, was that she did not possess a cognitively integrated ability to perform a salchow. But this invites the question: what would appropriate cognitive integration involve, at the group level? The anti-intellectualist who embraces non-summative knowledge-how will need to provide an illuminating answer.
Collective epistemology has burgeoned in recent years,26 , and with it, we expect the question of collective knowledge-how will be a topic of increased attention in epistemology, much as the topic of collective propositional knowledge has received increased critical focus. The foregoing is just a rough sample of some of the questions that we think should be at the top of the agenda for those who inclined to take group knowledge-how beyond the level of metaphor.
8.3 Conclusion
We are at an exciting time for research on practical knowledge. Much contemporary epistemology has been driven by interest in propositional knowledge which was driven by responses to the Gettier problem. Research on knowledge-how promises to recover an ancient tradition of thinking about a theory of knowledge and a theory of conduct as two sides of the same coin. Moreover, recent work on extended and group knowledge promises to extend our conception of the nature of knowledge-how beyond a standard intracranical understanding of human cognition.
8.4 Further reading
• BonJour, L. (1980). Externalist theories of empirical knowledge. Midwest Studies in Philosophy, 5(1):53–73
• Carter, J. A. and Czarnecki, B. (2016). Extended knowledge-how. Erkenntnis, 81(2):259–73
• Clark, A. and Chalmers, D. J. (1998). The extended mind. Analysis, 58(1):7–19
• Lackey, J. (2016). What is justified group belief? Philosophical Review, 125(3):341–96
• Palermos, O. and Tollefsen, D. (Forthcoming). Group know-how. In Carter,
J. A., Clark, A., Kallestrup, J., Palermos, S. O., and Pritchard, D., editors, Socially Extended Epistemology. Oxford: Oxford University Press
• Poston. (2008). Internalism and externalism in epistemology. Internet Encyclopedia of Philosophy. http://www.iep.utm.edu/int-ext
8.5 Study questions
1 What is the difference between knowledge-how and knacks?
2 What do Clark and Chalmers think the parity principle should lead us to say about the cases of Inga and Otto?
3 Is the ‘ATM’ case a case of extended knowledge-how? Explain.
4 What is the extended luck problem?
5 What is Palermos and Tollefson’s assessment of the Corvette case?