Custom Search

Saturday, 4 July 2009

The Ship of Theseus

“The Reconstructed ship is not Theseus’s ship, but would have been if the planks of Theseus’s ship had not been replaced.” Is this position coherent?

Coherent or not, this is the view espoused by E.J. Lowe as a solution to the famous paradox, drawn out by Thomas Hobbes, from the writings of Plutarch. For the purposes of this short essay, I will presume that the reader is already aware of the paradox wherein the original ship; Theseus’s ship is gradually repaired. Planks are replaced gradually whilst those removed from it are reassembled, eventually producing both a renovated ship and a reconstructed ship which seemingly both have claims to be one and the same as (or numerically identical with) Theseus’s ship. If this puzzle is unfamiliar then I refer the reader to Lowe’s ‘A Survey of Metaphysics’ pages 25-28 where he states the paradox more thoroughly.

Lowe’s solution is one of several solutions proposed and I believe that because it attempts to posit an objective fact-of-the-matter regarding identity, it quickly falls into contradiction. Lowe is not alone in making this error; indeed many critics of his view also make this mistake. I believe it is a mistake to regard identity as something objectively real; something that inheres in things, rather than as a concept that we as rational beings bring to bear upon the world. To regard identity as something intrinsic, rather than conceptual, necessitates contradiction.

I hope in this essay to demonstrate that Lowe and others who posit an objective notion of numerical identity have overlooked something fundamental about the nature of identity and perhaps reality as a whole. It is my view that a transcendental idealist perspective offers profound insight into the nature of identity, and that any view claiming either ship to actually be Theseus’s ship can only ever be partly correct, until the role of subject and context are given their due status in addressing the puzzle.

Kant’s notion of an antinomy, wherein reason leads us inexorably towards two opposite and contradictory views, seems to capture the very essence of the problem as resulting from our ways of thinking about things, rather than it (the paradox) highlighting anything problematic about things, such as ships, as they are in themselves. The solution also lies in the structures of our thoughts about things; this is why no philosopher feels the need to perform a practical experiment, or travel to Athens to better understand the problem.

Essentially all that Lowe is claiming, is that in cases where form constancy and matter constancy conflict, form constancy takes priority over matter constancy; where there is no conflict either criterion has the potential to be adequate criterion for numerical identity. He does this by proposing subjective criteria for what qualifies as a part of something and what does not. For example (italics are added for the sake of emphasis):

“…we must decide then to which ship the various parts belong at the time in question.” (Lowe, E.J. 2002, p.31)

or :

“Once the original parts of a ship have been appropriated by another, distinct ship, I suggest, they cease to be parts of the original ship:” (Ibid. p.31)

Lowe’s solution rests entirely upon agreement with this principle and not upon anything more concrete than that. These criteria are in themselves problematic. In his paper on the Ship of Theseus, Theodore Scaltsas lists eight criteria for what might qualify as a part of something:

“x is a part of y or of z if

(1) x is an attached and functioning part of y;
(2) x is an attached but non-functioning part of y;
(3) x has been removed from y but is not used as a part of any other object or for any other function;
(4) x has been removed from y but it has not been replaced by any other such part;
(5) x is a part that could be fitted to y but it has not been fitted yet[…];
(6) x is one of the original parts of y (when y was constructed);
(7) x is originally a part of y (in x’s first use as a part);
(8) x has at some point in its life span been functionally attached as a part to y;”

(Scaltsas, T. 1980, p.155)

His list is by no means exhaustive, nor is it intended to be. As he says:

“I am convinced that this is an incomplete list of the various senses in which x can be taken to be a part of y. Even now it would take a computer to figure out the possible conflicts of ‘being a part’ sufficiency conditions that can occur in various cases in which two or more of the above criteria would oppose one another;”
(Scaltsas, T. 1980, p.155-156)

The problem arises that it would be hard to argue for a hierarchy of such sufficiency conditions. There are numerous ideas we hold about what qualifies as a part of something, and in cases where y and z (or the reconstructed ship and the renovated ship) both have a claim on x we have no objective means of deciding which condition takes priority. This is precisely where the objective account fails.

For Lowe’s answer to have any objective validity, it must be self-evident to anyone looking at the situation regardless of any subjective opinions; yet, several times, Lowe makes appeals for the reader to form opinions about what constitutes a part of a ship and whether it remains so or not when removed, replaced and so on. It falls to each individual to decide whether form constancy or matter constancy have the greater bearing in such cases; when disagreement about these things arises, neither side can bring evidence to settle the matter decisively. To quote Scaltsas once more:

“There is simply the possibility of being differently inclined on the subject of condition hierarchy from your fellow-men, and there is no evidence they can bring to you or you to them to convince one another who is right.”
(Scaltsas, T, 1980, p154)

At least one person can object on perfectly solid grounds to Lowe’s claim that form constancy is more important for numerical identity. Consider the Archaeologist who wants to confirm the age of Theseus’s ship. Lowe cannot rightly say that he should go to the harbour to retrieve planks for carbon dating. In such a case, regardless of whether or not the removed timbers have been reassembled or not, our archaeologist should go to the museum. Lowe would be quite right to point a sailor to the harbour but what is underlined by this, is an issue of context that is overlooked in any solution that posits a fact-of-the-matter answer. How could either the sailor or architect ever agree which ship is numerically identical with Theseus’s? Both prioritise different sufficiency conditions for numerical identity; neither is able to offer the other the least bit of evidence, or proof, of the superiority of either view.

Another common objection to Lowe’s answer is that it violates “the only X and Y principle”. In brief, this principle states that whether X and Y are identical or not can only depend on X and Y, and not upon the existence (or lack thereof) of something else. In his solution, Lowe gives his reasons why he feels this is not the case, on the premise that in the two situations the reconstructed ship would be quite a different ship. To apply Scaltsas criteria, Lowe claims that in the no-renovation situation, conditions (3) and (4) obtain for continued part identity; conversely, in the renovation situation they do not. Therefore, in the no-renovation case the reconstructed ship is Theseus’s ship. In the renovation case, the renovated ship can claim numerical identity on the basis that conditions (1) and (5) obtain with regard to its parts, but problematically for Lowe the reconstructed ship can claim that conditions (3) and (6) obtain in the case of its parts. As demonstrated above he cannot claim either ship to decisively be the original in the renovation situation.

What qualitative difference there is, is indeed puzzling to me. The parts of the reconstructed ship in both cases are qualitatively identical, so how the difference in numerical identity of the whole obtains, is problematic for Lowe. It is hard to see what objectively is different about the reconstructed ship in the two cases. On the other hand, the concepts applied to the parts in the two situations are different. It is perhaps in this regard that Lowe falls foul of the “only X and Y principle”. Because concepts of part-hood are not features of things in themselves, there is no objective way to settle the issue of priority of sufficiency conditions; the subject determines whether X is the same as Y, in the same way as the subject decides which ship the parts are a part of. The only X and Y principle as stated above requires an objective fact-of-the-matter answer, that does not depend upon anything other than the original ship and the reconstructed or renovated ship for its validity.

In attempting to show that his solution does not violate the “only X and Y principle” Lowe clearly indicates his acceptance of its validity whilst basing his argument on concepts that are by their very nature, independent of either X or Y. This is not coherent outside of a more subjective framework; one that already rejects the “only X and Y principle” as it has been stated. The most that Lowe can hope to claim (without establishing an objectively valid hierarchy of part sufficiency conditions) is that, depending upon one’s views about the hierarchy of sufficiency conditions for parthood; there is a case for claiming either ship to be numerically identical with the original. Furthermore the subject plays a crucial role in determining the continuity of identity and the “only X and Y principle” needs to be modified to take account of this.

It would be entirely plausible to claim that, in strict terms, neither ship is objectively numerically identical with Theseus’s Ship. Numerical identity, as seen, depends upon our opinions as subjects and not entirely upon objective properties of things. As rational beings we can consent to either ship being regarded as Theseus’s ship, without it having to actually be the original ship; in much the same way as we accept that the new actor is still James Bond, without him having to be Sean Connery. Since both the renovated and reconstructed ship are merely regarded as Theseus’s ship (as opposed to actually being so,) and both, with commonly accepted (though not objectively valid) reasons, it is no longer problematic that there are now two ships. Neither ship is Theseus’s ship. Both represent the continuance of different concepts we have about numerical identity, but because identity is something we get to decide about, neither ship can be numerically identical in any more concrete sense than subjectively. Lowe’s solution may be coherent within his conceptual framework but, the argument he uses is symptomatic of the misunderstanding that gave rise to the paradox; that numerical identity is purely objective and that we discover it rather than determine it.


Lowe, E.J. (2002) “A Survey of Metaphysics”, Oxford University Press,

Noonan, H. W. (1985) “The Only X and Y Principle” Analysis, Vol. 45, No. 2, pp. 79-83, Blackwell Publishing [Online] Available from: [Accessed: 11/01/2009 02:40]

Scaltsas, T. (1980) “The Ship of Theseus” Analysis, Vol. 40, No. 3 pp. 152-157, Oxford University Press [Online] Available from: [Accessed 11/01/2009]

No comments:

Post a Comment

keep it nice, or funny...