Philosophy, Film, Politics, Etc.

Friday, November 1, 2013

Knowledge and Skill: the intersection of science, philosophy and common sense


Can you become a great martial artist without breaking a sweat?  In this classic scene from the 1999 sci-fi film, The Matrix, Neo (played by Keanu Reeves) seems to learn Kung Fu by having the information uploaded directly to his brain.

If you cannot view the video, here's what we see:  Computer monitors show simple, 3D images of a brain and a martial arts position.  The computer is plugged into Neo, who is strapped to a chair with his eyes closed.  Neo opens his eyes with a noticable exhale, blinks, turns to Morpheus (played by Laurence Fishburne), and says, "I know Kung Fu."  Morpheus leans in, intrigued, and replies, "Show me."

Morpheus is not clearly convinced.  Perhaps he is not sure the upload was successful.  Or maybe Morpheus believes that the upload was successful, but that it was only part of the learning process.  The skill has not really been attained until it has been demonstrated.

It doesn't matter what Morpheus thought.  The question is, what should we think?  Could Neo become a Kung Fu master without ever demonstrating the ability?

To be a great martial artist is to have great skill and knowledge, but these might not be the same thing.   So perhaps there are two distinct questions we could ask.  The original question was:  Does Neo know Kung Fu before he demonstrates it?  The second question is:  Is Neo skilled at Kung Fu before he demonstrates it?

Since this is science fiction, perhaps the most respectable answer to these questions is maybe. However, I will argue that the best available answer to both questions is probably not.  Even if our answer is not definitive, exploring the questions could help us sort out some confusions in our understanding of skill and knowledge.  In particular, I will try to sort out some confusions at the intersection of three areas of discourse:  common sense, cognitive science and epistemology.  In the process, I will critically consider some recent attempts to do the same.


The first step is to consider ways of distinguishing between skills and knowledge.  For example, it is obvious that Neo cannot be skilled at Kung Fu if he does not know Kung Fu, but there may be ways he can know it without being skilled at it.

In common sense, a person can know about Kung Fu without knowing it in the sense required for mastery of a skill.  One can read books about Kung Fu, for example, and learn all about the different moves and principles behind it, but if one does not learn how to perform the moves and apply the principles in practice, then we would say they lack the know how.  This is the common distinction between practical and theoretical knowledge.  Theoretical knowledge of a skill is what you know by reading books about a skill.  Practical knowledge is knowledge demonstrated in the exercise of a skill.

The practical/theoretical distinction can be misleading and deserves closer inspection.  After all, we expect experts to be able to apply the principles we can read about in books.  So their practical knowledge can include the theoretical principles contained in books.  Furthermore, we might suppose that every aspect of Kung Fu can be subject to theoretical analysis:  Any part of the Kung Fu master's skill can be the subject of theoretical interests.  It might therefore look like practical knowledge contains theoretical knowledge while theoretical knowledge contains practical knowledge, and there is nothing which limits one with respect to the other.  In that case, we should think of them as the same.  However, this is a mistake.  All we have seen is that a person can have both practical and theoretical knowledge of the same aspects of a skill.  That does not mean the practical knowledge and theoretical knowledge are the same.  They just have the same object.  Practical knowledge and theoretical knowledge are different ways of knowing the same thing.

If you understand the principles of Kung Fu theoretically, but not practically, it means you can talk about Kung Fu, and even analyze Kung Fu, without being able to demonstrate them in practice.  If you understand the principles practically, but not theoretically, it means you can demonstrate them without being able to talk about them, or without being able to analyze them as such.  A Kung Fu master has exemplary practical knowledge of Kung Fu, but can be more limited with respect to theory.  Conversely, a great teacher has exemplary theoretical knowledge, but can be more limited with respect to practice.

For the present purposes, it doesn't matter if Neo's upload gives him much theoretical knowledge of Kung Fu.  When Neo says, "I know Kung Fu," Morpheus wants a demonstration of practical knowledge.  He does not invite a discussion of its principles.  Thus, the original question can be rephrased:  Does Neo have practical knowledge of Kung Fu before he has demonstrated it?  

The more general question is:  Can a person gain practical knowledge of a skill without actually practicing the skill?


Cognitive scientists talk about skill in terms of procedural memory (also called "unconscious" or "implicit" memory).  This is in contrast to declarative (i.e., "conscious" or "explicit") memory.  The idea is that bodily movements are learned differently than intellectual functions.  Factual knowledge--knowledge which underlies conscious reflection, theoretical analysis and discourse--is neurologically distinct from motor skills. Scientists recognize that declarative and procedural knowledge work together.  For example, in a 2010 paper, Shumita Roy and Norman W. Park argue that tool use depends on declarative as well as procedural knowledge.

This approach has been criticized in a paper by Jason Stanley (Department of Philosophy, Yale) and  John W. Krakauer (Department of Neurology and Neuroscience, Johns Hopkins) published earlier this year in a journal of neuroscience, Frontiers.   (They present their argument in a slightly more accessible format in this New York Times column.) According to Stanley and Krakauer, scientists have been looking at the relationship between knowledge and skill the wrong way.  They argue that there are cultural prejudices at play which have led to scientific confusion.

Stanley and Krakauer want to use concepts and arguments from epistemology to sort out perceived problems in cognitive neuroscience.  They claim that their results have implications for our common sense understanding of the distinction between practical and theoretical knowledge.  One of their ideas is that practical knowledge is every bit as intellectual as theoretical knowledge.  They deny that there is any basis for the distinction between practical and theoretical knowledge at all, if we are to understand that distinction as a difference between two kinds of knowledge.  They reject the idea that motor skill is just procedural knowledge, and deny that so-called procedural knowledge is knowledge at all.  They say that motor skill depends on knowledge and other factors which sustain practical ability.

A substantial part of their argument is about vocabulary.  They do not like the way cognitive scientists use the words "knowledge" and "skill."  They argue that epistemology offers a more conceptually grounded and reasonable vocabulary.  However, they are not only talking about vocabulary.  They also claim to be challenging some cultural prejudices about intellectual capacities.  They suggest that the dominant paradigm leads scientists to unfairly devalue some kinds of knowledge (for example, the knowledge required for mastery of sports) and unfairly elevate other kinds of knowledge (namely, the knowledge more generally associated with academia, such as knowledge which relies on the precise acquisition of specialized speech patterns.)  People with book smarts are no more intellectual than star athletes, they say, and it is only a cultural prejudice which accounts for the popular contrary opinion.  All knowledge is intellectual, they say, and so they call themselves Intellectualists.

I suppose there is a popular prejudice against the intellectual capacities of athletes, though it has been challenged many times and from many quarters over the years. I'm not sure how widespread it is.  More importantly, I am suspicious of the idea that it has had such dire effects on cognitive science.


Let's consider Neo's situation in a little more detail.  I will retain the vocabulary from cognitive science for the time being (keeping Stanley and Krakauer's reservations in mind).  Neo's upload affects his procedural and declarative memory.  If this were otherwise, then Morpheus would not ask him for a demonstration at all.  Neo is expected to demonstrate some competence at martial arts, and not merely demonstrate the ability to talk about or analyze it.  However, Neo's muscles and joints are not in shape yet.  Practice does not just train the brain.  It trains the body.  So we should not expect Neo to perform at a master level until he has practiced a bit.  By demonstrating his knowledge, he gets in shape.  By getting in shape, he improves his practical ability.

So we have  three elements to consider:  knowledge, skill and practical ability.  These may be three distinct elements, or perhaps they are different ways of talking about the same thing.

On one view (or family of views), which I'll call Intellectualism, Neo's upload can provide him with all of the knowledge required for Kung Fu, even though he does not have the practical ability.  The assumption is that he knows Kung Fu by virtue of the information in his brain.  We can further distinguish between two versions of this view.  According to one version, Neo's skill does not improve upon demonstration of his ability.  Knowledge and skill are identical, and not dependent upon practical ability.  Let's call this thesis Strong Intellectualism.  According to the other version, Neo's skill does improve upon demonstration of his ability . Let's call this thesis Weak Intellectualism.  According to Strong Intellectualism, knowledge is sufficient for mastery of a skill.  According to Weak Intellectualism, knowledge is not sufficient for mastery of a skill.  The Weak Intellectualist says skill is a combination of knowledge plus practical ability. Yet the Weak and Strong Intellectualists agree that mastery of a skill does not require knowledge gained through practice.  The Weak Intellectualist does think practice is necessary, but not to improve knowledge--only to improve skill.

Against Intellectualism, there is the view that the practical abilities gained through the demonstration of Kung Fu are necessary to give Neo the knowledge sufficient to be skilled at the craft.  On the Anti-Intellectualist view, the upload can only reliably give Neo some of the knowledge required for the skill, and the demonstration is necessary to supply him with additional knowledge.


I favor Anti-Intellectualism.  Before I make a positive argument for it, however, I will point out some problems with Stanley and Krakauer's argument for Intellectualism. Stanley and Krakauer are Weak Intellectualists.  They recognize the empirical evidence for so-called "procedural knowledge," and they recognize that this is required for skill.  Yet, they deny that this is knowledge.  They claim it is a non-propositional component of a skill.  Since it is non-propositional, it cannot be knowledge.  This follows from their explicit assumption that knowledge is, minimally, a relation with propositional content, though they do not say what that entails.

The idea of propositions underlies much philosophical work since Frege, Russell & Wittgenstein, but its meaning remains controversial and vague.  On some readings, propositions are possible worlds.  On others, they are structured components of worlds.  On others, they are whatever can be true or false.  All of these interpretations remain problematic.  While the concept of propositions is often taken for granted in many philosophical discussions, its foundations and coherence remain heavily questioned.  If they were publishing in a philosophy journal, we should not be surprised that Stanley and Krakauer do not give a definition of what "propositional content" entails.  However, they are publishing in a neuroscience journal.  When a problematic and controversial philosophical concept is used to criticize the interpretation of scientific experiments in a scientific journal, we should pause and wonder what is going on.

Let's look at how this plays out in their New York Times article.  They quote Suzanne Corkin, a neuroscientist writing in a newspaper article for The Guardian, as follows:

Henry was not capable of learning new information, though his knowledge of past events — the Wall Street Crash, Pearl Harbor and so on — was clear. Only a very few tiny details of TV programmes he watched repetitively ever stuck. He could, however, learn and retain new motor skills, which led to important understanding of the difference between conscious memory and unconscious. The latter category would include learning how to play tennis or ride a bicycle, or even play the piano — things that the brain encodes and transmits to the muscles through conditioning, memories which we come to think of as intuitive.
Here's how I would interpret the main points of this passage:  Henry (aka H.M.) was impaired with respect to his declarative memory.  Yet, H.M. could improve his motor skills.  Here is how Stanley and Krakauer interpret the passage from Corkin:
According to [Corkin], H.M. was able to “learn and retain new motor skills” (and even improve). Examples of such learning are “how to play tennis or ride a bicycle.” H.M. is therefore taken to show that motor skills, a paradigm example of which is tennis, are not the employment of knowledge.
Did Corkin really say that the exercise of motor skills is not an employment of knowledge?  Stanley and Krakauer will surely acknowledge that scientists are apt to say that it is "procedural knowledge," but they claim that this is a contradiction in terms.  They say that it is not really knowledge, because it has no propositional content.  But what is propositional content?  They haven't said.  Can procedural knowledge have propositional content?  I think Corkin can be ambivalent about the answer to that question, since it does not seem to have any empirical basis in the first place.  Corkin is not saying that motor skills are not an employment of knowledge.  She's saying they're not an employment of declarative knowledge.  But tennis, for example, surely requires both procedural and declarative knowledge.  (It would be extremely uncharitable to take Corkin to mean that a person could learn how to play tennis without being conscious, or without having any conscious memories of the sport.)  Tennis is not a motor skill, strictly speaking, but a skill which involves both motor skills (procedural knowledge) and intellectual skills (declarative knowledge). And there is no scientific reason to think that either of these are (or are not) propositional, because we have no empirical understanding of what that even means.  So why do Stanley and Krakauer claim that procedural knowledge is not really knowledge?

Stanley and Krakauer's argument relies on the claim that procedural knowledge is not propositional knowledge, because it is not knowledge of facts. But they have no coherent argument for this.  They point out that procedural knowledge is defined in opposition to declarative knowledge, where "declarative knowledge" is defined as explicit knowledge of facts; however, they reject the category of "declarative knowledge" as being too poorly defined.  They insist instead that knowledge does not have to be explicit or capable of articulation.  Therefore, the fact that procedural memory is not declarative knowledge does not mean that procedural memory is not knowledge in Stanley and Krakauer's sense of "knowledge."  Instead of claiming that procedural knowledge is not knowledge, they could just as easily claim that it is implicit knowledge of facts.  They have no reason to claim that procedural memory is not propositional knowledge.

Let's look at how Stanley and Krakauer respond to another neuroscientific development.  They criticizie Roy and Park, who we have already seen have distinguished between tool use and motor skills.  They say that Roy and Park's conclusion is bizarre and confused.  Stanley and Krakauer want motor skill to be associated with both procedural and declarative memory, remember.  They want to upset the whole procedural/declarative distinction.  Here's how Stanley and Krakauer criticize Roy and Park:
the obvious alternative — that motor skill, like any other cognitive task, requires knowledge. Perhaps patients with H.M.’s condition can acquire one part of what is required for skill, but not other parts. If so, then neuroscientists have made an overly hasty identification of a noncognitive component of motor skill with the category of motor skill itself. This identification overlooks the knowledge component of anything that common sense classifies as a motor skill, like tennis.
The argument is this:  Why claim that D.M. has all the right motor skills?  Why not claim that what he's missing--the declarative memories he cannot retain--are also motor skills?  They say that neuroscientists are mistaking one aspect of motor skill (which scientists call "procedural memory") for the whole.  It rather seems that they are conflating the common-sense notion of skills (many of which include both declarative and procedural knowledge) with the scientific notion of motor skills (which are limited to procedural knowledge).

This might just be a terminological dispute, but Stanley and Krakauer claim to be proposing a radical shift in the way scientists go about framing and interpreting their experiments.  This, however, is not so clear.  Their argument is that there is a cultural prejudice at work and that it is leading to social inequity. They write, "A skilled archer knows what to do to initiate the activity; this is in part why she can decide to do that activity. Still, we are not supposed to call LeBron James a “genius” because cultural biases have infected science without the moderating input of the humanities."

Is that really so?  There is no evidence that any of the scientists Stanley and Krakauer have cited would argue that LeBron James, or any other athlete, should not be considered a genius.  There's no evidence that their scientific arguments support the prejudiced view that people with highly developed practical skills are less intelligent than people with "book smarts."  Stanley and Krakauer's argument is that this anti-practical bias is based on the mistaken idea that procedural knowledge does not require intelligence, and that declarative knowledge is the only true sign of intelligence.  Perhaps many people do hold such a view, but it is hard to see how Corkin or Roy and Park can be found guilty of supporting it.

Stanley and Krakauer have not shown evidence of any significant prejudice or confusion in the scientific literature.  They have only introduced a problematic vocabulary from epistemology and effectively confused the science of skill.  If Stanley and Krakauer's paper and Times article demonstrate anything, it may just be that we should not confuse the vocabularies of cognitive science and epistemology.

None of this counts as an argument against Intellectualism, though it does undermine Stanley and Krakauer's argument for it.  Now I will turn to a more general argument against Intellectualism.


Neo, like everybody else, has a neurologically encoded representation of his own body.  Neuroscientists discuss such representations when they explain, for example, the perception of phantom limbs.  When our self-representation does not match the physical reality of our body, strange things happen.  We can hallucinate in alarming ways.

Prior to the upload, Neo's physical body is not used to the motions of Kung Fu.  His self-representation is not that of a Kung Fu master.  After the upload, he thinks of himself as a Kung Fu master, though the only thing that has changed is his brain.  The body that his brain represents has not changed.  The question, then, is this:  Is his subconscious, neurological self-representation that of a Kung Fu Master?  If it is, then it does not match the reality of his body.  The cognitive content of his self-representation would be false.  We should expect this to have alarming consequences for Neo.  He would not act the way he thinks he should.  He might find it very difficult to stand up, let alone perform martial arts.

For the upload to work, Neo's self-representation cannot be markedly different from what it was before.  When Neo begins to demonstrate his know how, his body changes and, in turn, his self-representation changes as well.  In this way, his procedural memory develops.  Only by practicing Kung Fu can Neo develop an internal representation of his body which is consistent with the way his body actually moves when he does Kung Fu.  Only after he has done it can we say he knows how to do it.  Only then does he have the knowledge.  And only then does he have the skill.  Since he gains the knowledge at the same time, and by virtue of the same processes, we can say (along with the Strong Intellectualist) that knowledge is sufficient for skill.  Unlike the Strong and Weak Intellectualists, however, we have empirical evidence suggesting that practical ability is also part of knowledge.

Stanley and Krakauer might insist that Neo is not gaining new knowledge by developing his procedural memories in this way.  However, that seems very counter-intuitive.  Neo is developing new representational structures in the form of memories which engender new, reliable behaviors.  That seems to be just the sort of thing people mean when they talk about knowledge.  Perhaps there is some specialized sense of "knowledge" which discounts procedural memory, but Stanley and Krakauer have not given a coherent argument to that effect.  Common sense and cognitive science seem to agree that procedural memory counts as knowledge, and the example of Neo learning Kung Fu demonstrates why.  By developing his practical abilities, Neo learns.  Learning implies knowledge.

I have not taken a position on whether or not procedural memory is propositional.  I have only claimed that it is knowledge, according to common sense and cognitive science.  If it is non-propositional knowledge, so be it.  If it is propositional knowledge, so be it.  I don't see any need to take a position on that issue.  Until the notion of "propositional content" has some empirical significance, it should not influence the scientific study of skill.  It is enough to observe that there are empirically sound reasons for distinguishing between procedural and declarative memory, and we have every reason to think of these as varieties of knowledge.  Skills rely on both to varying degrees.


So, can Neo become a master of Kung Fu by upload alone?  Since this is science fiction, perhaps we should stick with "maybe," but based on what we know, I think the most plausible answer is no.  Morpheus was right to ask for a demonstration, though perhaps not for the reason he thought.  If my analysis is accurate, then Neo could not know how he could do Kung Fu unless he had reliable and accurate procedural memories of doing Kung Fu.  He could not have the skill or the knowledge until he performed.

Stanley and Krakauer disagree, but their argument is based on a conflation of vocabularies.  They want us to believe that knowing how to do martial arts is just part of the skill (perhaps it is the part that can be learned discursively, though I don't think they're clear on this).  The other part of the skill, which relies on procedural memory, is not knowledge at all.  Yet, they do not offer a coherent argument for this position.  Procedural memory appears to be a part of knowledge or even a distinct a kind of knowledge, propositional or not.

I agree with Stanley and Krakauer that athletics and academics are both capable of demonstrating intelligence.  I do not think a great scientist is necessarily smarter than a great athlete.  However, I think they may be smarter along different dimensions, and it makes sense to consider one physical and the other intellectual.  The reach of intellectual intelligence is probably wider than the reach of physical intelligence.  It is true that both physical and intellectual intelligence are involved in both athletics and academics.  However, the types of skills required of an academic are not entirely contained in the skills required of an athlete, and vice versa.  There may not be a clear distinction between intellectual and physical skills--they may blend together, but they are distinct sets.  Also, I wouldn't assume that all skills are either intellectual or physical.  Some can be both, others can be neither.  Furthermore, there can be biological (and that includes developmental) reasons why some people are better at some skills than at others.

In the end, we should not be convinced by Stanley and Krakauer's argument against the distinction between practical and theoretical knowledge, or between intellectual and non-intellectual pursuits.  While athletics certainly requires intelligence, and while it can certainly involve intellectual pursuits (studying strategy, for example), the term "intellectual" is useful for indicating a different skill set.  Stanley and Krakauer should recognize that some athletes are more learned when it comes to strategy, while others are more skilled when it comes to execution.  Strategy is considered more intellectual, because it can be learned discursively--through spoken and written language.  Strategy depends more on declarative memory, whereas execution depends more on procedural memory--on the movement of the body.  Thus, a person can have advanced theoretical knowledge of a sport or martial art without having the knowledge required to execute it with any skill.  The practical and the theoretical are two different relations to the same things.  Furthermore, practical ability is part of knowledge.  We do not have skills without practical ability, and we do not have the knowledge required for the mastery of a skill without having mastery of the skill itself.  For that reason, we should reject both Weak and Strong Intellectualism.