That Tug Boat Captain, He Got Nerves of Steel

That tug boat captain, he got nerves of steel.
That lock ain't no wider than the eye of a needle.
Six coal barges ain't no fine cotton thread,
And the muddy river bottom ain't no box spring bed.

Put on your shoes honey go grab your coat.
We'll walk up to the river watch them old tug boats.
I cannot tell you why love grows cold,
Anymore than I can tell you where they're taking that coal.

Mike West, Tug Boat

TITLE?

Isak Dinesen once said that the many paths to truth include those going through Bourgogne. The distilled beverages of the departments Charente and Charente Maritimeshould inspire similar reverie.

While the seven officialsubdivisions of the Cognac region were delimited by law in 1936 (eighty-one years after the classification of the Bordeaux), two influential earlier surveys from were used.[1] By 1918, date of the previous surveys, the sciences of oenology and geology had advanced enough to commission one survey based entirely on soil conditions. The other survey was done in the traditional manner, by a committee of expert testers. Amazingly, the two maps agreed in many ways. For example, the geologists determined that (what would be designated in 1936 as) the premier cru Grande Champagne region and the deuxième cru Petit Champagne regions reliably had the most chalky subsoils. The tasters determined that cognac produced in these two regions were the best.

Philosophers well versed in “Frege’s problem”[2] will already appreciate some of the philosophical interest here. Consider the two definite descriptions, “the area in the departments Charente and Charente Maritime that allow gifted farmer/distillers[3] to produce the best Brandy,” and “the area in the departments Charente and Charente Maritime with the chalkiest subsoil.” Now consider a person who believes truly that the premier cru cognac comes from this area. The truth of (de dicto) attribution of this belief will depend upon how the area is described. For example, it would be false to say of a geologically naive taster that she believes that the departments Charente and Charente Maritime with the chalkiest subsoil allow production of the greatest cognac.

The attempt to delineate a plausible logical semantics for such attitude ascriptions is probably the most difficult issue in the philosophy of language. Strangely, epistemologists have largely been silent about this. Given the orthodox view that belief in a proposition is a necessary condition for knowledge of that proposition, one would expect a robust epistemological literature around this issue.

To see the relevance of this, suppose that our geologically ignorant tester is taught that the correct French to English translation of “Je sais que ce cognac est très bon.,” is, “I know that this soil is very chalky.” Would the English language assertion be correct? Does she really know that the soil is very chalky? The claim is both true and justified,[4] but most of us would conclude that the belief expressed by the proposition cannot for her play the proper role in providing knowledge. Therefore she can’t be said to know that the soil is very chalky.

In what follows we hope to show that such examples undermine both epistemic internalism and externalism. By showing how each view’s problems are fruitfully thought of as two sides of the same coin, we are able to motivate a new position, where (externalistically) knowers are not required to have cognitive access to the justification of a proposition, but (internalistically) they are required to access the content of that proposition. We conclude by explicating some interesting consequences of the new position.

I. THESIS: THE SOPHISTICATS

Here is a cartoon version of epistemic internalism. A belief is only justified if the justification of that belief is directly cognitively accessible. By this we mean that with minimal prompting the viewer can produce the justification in proof-like form.

In his recent (2005) overview of the debate between epistemic internalists and externalists, Laurence Bonjour divides objections to internalism into two forms: (1) unsophisticated epistemic subjects, and (2) the threat of skepticism.[5]One argues for the first by noting that internalism does not correctly describe the epistemic abilities of children and higher animals such as parrots, dolphins, elephants, apes, and perhaps dogs. Arguments of the second type are in a sense extensions of these arguments, but instead of the unsophisticated they focus on everybody. Here it is argued that if internalism is true then nobody can have knowledge, because internalism requires too much.

The kind of counterexample we would like to discuss cuts across this division. Interestingly, if we examine the most sophisticated epistemic subjects we see that expertise always requires a particular lack of sophistication. Consider the tugboat captain chronicled by the narrator of Mike West’s song of the same title. Contemporary tugboat captains have recourse to Globalized Positioning System, radar, and up-to-date information about the bodies of water they traverse to accomplish their difficult task. However, like their pre GPS ancestors, they still need the ability to assess water conditions by look alone. The nearly impossible task of maneuvering a barge under a bridge or through a lock requires an ability to discern the status of the often violent underwater currents on rivers like the Mississippi.

Strangely, tug captains that possess a wealth of propositional knowledge about their boat and the river are inarticulate about how eyeing the river allows them to instinctively know how to guide the boat without crashing into a bridge.[6] Thus, a tugboat captain can know the proposition, “To avoid smashing into the bridge, one must turn starboard now,” while having no grasp of the canonical justification for this.

Of course theepistemic internalist has a variety of responses, but prior to assaying any of them, we do well to draw out the proper moral, which is that cartoon internalism spectacularly fails when practical abilities about which we are inarticulate play a role in justification.

If this is the correct analysis of the problem then the internalist has a simple response. A more sophisticated (daguerreotype?) version of internalism would hold that the knower needn’t actually know the canonical justification for her claim, but just be (in some suitably restrained way)[7] able to get herself in a position to know the justification. From the idealization, we have that though the tugboat captain has no idea what perceptual cues are relevant and how her mental algorithm manages those cues to issue in judgments, she can still be said to have the propositional knowledge. For with enough idealization she could get herself to know this.

This response is bad in a number of ways. First, with enough idealization my cat Homer can learn that Angus Young is AC/DC’s lead guitarist. Surely he doesn’t know that though. Second, it is unwarrantedhubris to assume that such procedures can be stated (in a proof-like and non-circular manner) for all such human capacities. Consider the old positivist distinction between discovery and justification.[8] By this view whatever actually goes on in a scientist’s head is irrelevant to justification, which is obtained by a scientific method which at least (by the standard deductive nomological model) algorithmic in the sense that the justifications can be mechanically checked (albeit perhaps not mechanically discovered).

Unfortunately, nobody has come up with such a procedure for solving the problems of high school physics textbooks, much less the actual problems scientists address. This was so intractable that in contemporary philosophy of science, the received wisdom now is that the non-linguaform practical capacities of the scientist in question are part of the justification of the propositions they believe. Part of what justifies technical and scientific claims is that people trained in the correct way reliably come to believe them.[9]

For our purposes the point is that if the tugboat captain or the scientist followed some procedure (in cases where they get it right) that could be specified with idealization, then it would presumably be possible to specify that procedure. If the procedure is mechanical, then one would expect that the question of whether the procedure is correctly followed or not could be checked by an algorithm. But every attempt to do discern a theory of scientific justification that satisfies these goals has been an embarrassing failure.

II. ANTI-THESIS

For Alvin Goldman[10] the common flaw in internalist theories of justification is the absence of a causal requirement:

Many of our counterexamples are ones in which the belief is caused in some strange or unacceptable way… In general, a strategy for defeating a noncausal principle of justifiedness is to find a case in which the principle antecedent is satisfied but the belief is caused by some faulty belief-forming process.[11]

Goldman holds that internalists have neglected this because such causal factors are not necessarily accessible to the agent’s first-person perspective. This leads him to account for justification in the following way.

The justificational status of a belief is a function of the reliability of the process or processes that cause it, where reliability consists in the tendency of a process to produce beliefs that are true rather than false.[12] Goldman’s reliabilism then holds that a belief’s justification depends on the causal process responsible for the belief to be a reliable indicator of truth.

Goldman’s assessment seems to harmonize with our common sense intuitions in a few crucial ways. First, it allows for degrees of justification where some processes are more reliable, varying less in producing true beliefs than other processes. This creates variant levels of justification. Second, it solves the unsophisticated knower problem, allowing us to attribute knowledge to some animals, children and untutored adults. One doesn’t have to have a reflective knowledge of why their belief is justified, the belief simply is justified by its causal process.

To see how implausible this is, consider Mr. Truetone,[13] a completely tone-deaf individual who has had a pitch-tracking device implanted in his brain. None of his cognitive architecture has been altered. He is given very precise note-beliefs by the device in the form of A, C#, Bb and so on. He is not able to identify what he hears as musical notes, and in fact is unable to engage in minimal conversation about melody or harmony. Instead, he can merely identify pitches by parroting what the device in his head tells him. He remains tone-deaf, displaying none of the discriminatory and discursive behavior of one who understands notes.

Suppose that one day he picks up the phone and says that the dial tone is an A sharp. Should he be justified in this belief? He is compelled to give this answer by the device, not by any comprehension of the dial tone. First, it should be clear that the agent with perfect pitch seems to have cognitive architecture that is different in some important way. While Mr. Truetone can give the correct public reading he doesn’t have the right subjective resources to understand his own claim. He is like the French speaking geologically naive cognac taster who is prompted to report, “I know that this soil is very chalky” when tasting cognac’s of a certain character.

III. SYNTHESIS

It is of more than passing interest that very similar (and not too counterfactual) gedankenexperiments undermine both epistemic internalism and externalism. This is a good key that we are hitting up against a symptom of a bigger problem, which we take to be a pervasive inability of modern epistemology to accommodate the role that practical abilities play in the justification of propositional knowledge.

The counterexamples are two sides of the same coin. Pace the internalist, we noted that in cases where people are inarticulate about the relevant practical capacities, yet none the less understood the proposition justified, they clearly have knowledge. Pace the externalist, we noted that in similar cases where the person does not understand the proposition justified, they do not have knowledge.

In the post Gettier tradition, the debate about epistemic accessibility has focused solely upon the rational assessment of beliefs. There seems to be a qualitatively different sort of accessibility in understanding.

The above examples draw out our intuitive notion of understanding the content of one’s beliefs that should accompany any claim of propositional knowledge. Mr. Truetone lacks the conceptual resources sufficient to understand his beliefs.[14] The internalist is right that cognitive accessibility is necessary for knowledge, and the externalist is right that the justification[15] needn’t be cognitively accessible.

IV. CONCLUDING THOUGHTS

If our argument is novel, it is probably because it seems to conflict with some redoubtable intuitions in contemporary philosophy: (1) one can know a proposition based on appeal to the testimony of authority,[16] and (2) that one can attribute de dicto beliefs to people who do not fully grasp the content of what they believe (semantic externalism).[17] However, reflection shows our view to be consistent with both of these trains.

Complex mathematical equations are simply beyond the comprehension of a large number of people. Should these people still be able to claim knowledge by deference to those who do understand? Laurence Bonjour seems to think it natural that we can claim to have mathematical knowledge by testimony, as do many others.[18] Mathematics gives a perfect example of claiming to believe a proposition without adequately understanding that proposition.[19]

Recently we were watching a special on John Nash who is currently doing research on some variety of fourth-order equations. We were completely unable to understand his explanation of his current research project. The only message that we got from that program was that, whatever he was doing, it was way over our head. Still, we are acquainted with the mathematical terms well enough to parrot his solution to some of these equations. Do we really know what we merely parrot after him? This is extraordinaerily implausible.

What this should reflect on testimony in general is that to know a proposition put forth by a reliable source, the content of that proposition must be accessible to the agent, in other words, relevant content must already exist in the mind of the agent. So consider another example of testimony: A reliable source tells me that my brother is in his room. This belief can count as knowledge because I have on many occasions seen my brother in his room and I understand what that proposition means. If on the other hand, a reliable source told me that Mr. Smithy, a name I have never heard, is in my brother’s room I cannot claim to know this because I lack the conceptual resources to understand who Mr. Smithy is. If this belief passed as knowledge it would entail a piece of ignorance, namely of the identity of Mr. Smithy, and that seems intuitively implausible.

On the other hand, if we understood the claims he was making, then it is consistent with our view that we know that the claims are true. So perhaps we do know that Fermat’s Last Theorem is true. The theorem is a simple piece of number theory, and easy to understand, even though the topology used in the proof is anything but.

Finally, if one really wanted to accord a positive epistemic status to the person who authoritatively parrots sentences that they don’t understand, one could say that the assertion is justified, but that the person asserting the claim does not possess knowledge.[20]

So we have seen that our account does accommodate robust intuitions about authority. However, it is partly based on such intuitions that semantic externalism is defended. Hilary Putnam noted that language communities practice a “linguistic division of labor,” where experts are in charge of knowing the causal, historical, and institutional facts that undergird identity judgements about meaning. As long as the rest of us defer appropriately to the experts, we can be said to have beliefs involving concepts of which we have very little grasp.

Ironically, it is just this aspect of semantic externalism that has engendered the most critical literature.[21] All we seek to contribute to this debate is the claim that our proposal for knowledge has nothing to do with it. Semantic externalism is about de dicto attribution of beliefs, not knowledge. Even (with the semantic externalist) if I could be said to have de dicto beliefs, the content of which I have no cognitive access, this does not entail that I have knowledge of the proposition in such cases.[22]

If we are correct that knowledge requires cognitive access to the content of a proposition, more issues are raised than solved. Even cursory examination of the debates concerning semantic internalism and externalism show that the theory of content is hardly in a more settled state than the theory of knowledge. If we are right, solution to one of the most difficult issues in epistemology requires a solution to one of the most difficult issues in the philosophy of language. Thus, things would go much simpler if we were wrong. However, we hope that the above arguments show that in this case we would cut ourselves with Occam’s razor.

REFERENCES