Paper on semantic externalism and psychological essentialism in lay speakers’ language use

Hi all,

me and my colleagues Henry Railo and Jussi Haukioja have an experimental paper on semantic externalism and psychological essentialism, now forthcoming in Philosophical Psychology. The paper presents data that supports Kripke-Putnam style externalism about natural kind term reference.

The paper can be found in whole here, but I describe it in some detail also in what follows.

According to a view in psychology of concepts called psychological essentialism, people believe that natural category members share some hidden, unobservable, empirically discoverable deep structure or essence, whose possession is necessary and sufficient for category membership. However, the psychological essentialists themselves don’t take their view to take a stand on what determines the extensions of natural kind terms.

We present two semantic adaptations of psychological essentialism, namely strict and hybrid externalistic essentialism. The strict view is motivated mainly by Kripke and Putnam, and it holds that speakers believe not only that natural category members share some hidden deep structure, but also that this hidden essence determines what belongs in the extension of the category term, even if the deep structure is unknown. The hybrid view, in turn, is motivated by some experimental studies suggesting that natural kind terms may be ambiguous between two senses, an externalistic and an epistemic (e.g. Braisby et al. 1996). In the externalistic sense the term applies to whatever shares the external essence in fact shared by the category members, but in the epistemic sense the term applies to whatever satisfies the identificatory criteria associated with the term. For instance, if all cats turned out to be robots while we believed them to be mammals, the term ‘cat’ would have referred to the robots in its externalistic, but not in its epistemic sense.

In the face of the proposed accounts, we examine Braisby’s et al. (1996) experiments which claim to undermine externalism. We argue that they fail because of both theoretical and experimental problems. We then go on to present two experiments of our own.We used scenarios that were presented to the participants in two stages. In the first stage the participants read descriptions of a natural kind, referred to by term t, and of some novel samples that either satisfied or not the identificatory information associated with t. Fictitious natural kinds were used. Here is an example:

A yellowish, bitter-smelling, fragile mineral called zircaum occurs widely in Mid-Siberian soil. Scientists generally believe that zircaum is the compound ACB.

In Northern Norway, a deposit of a substance just like zircaum is found – it is yellowish, bitter-smelling, fragile, etc. When scientists examine its deep structure, they find out that it is ACB, just like zircaum is believed to be. The scientists conclude that there is zircaum in Northern Norway.

We then presented the participant data about the original samples that were (by externalistic standards) to undermine the earlier judgement concerning the Norwegian sample:

A few weeks after the discovery in Northern Norway, scientists examine the Mid-Siberian substance more closely. Using methods and instruments more exact than previously available, they find out that they were wrong about the deep structure of the substance: the substance is KML instead of ACB. However, the substance found in Northern Norway was indeed ACB, just as the scientists thought it was.

After having read this passage, the participant was asked whether she considered the earlier judgement concerning the Norwegian sample (a) justified and (b) strictly speaking correct. Internalism predicts that the earlier judgement be considered strictly correct, as the sample fitted all of the identificatory criteria associated with the term at that time, whereas externalism entails that she consider the judgement false, as the Norwegian sample did not possess the real, external category essence. (I am here omitting some details and variations that we used in the scenarios.)

The results were that the majority of answers (69 %) were externalistic, and the minority (28 %) internalistic. Though the results undermine internalism, which cannot account for the externalistic answers, the results do not vindicate externalism. Instead, the results lend some support to the hybrid view, since it seems that the speakers used the category terms ambiguously between an externalistic and epistemic sense. However, we couldn’t rule out that the response pattern was due to the participants’ guessing or answering randomly, as they had no choice but to answer either internalistically or externalistically; a guessing strategy would have produced the same response pattern as the one predicted by the hybrid view.

Because of this problem, another experiment was conducted, where the participants were given the chance to pick one out of four alternatives with respect to the correctness of the earlier judgement. These were (i) that the judgement was correct, (ii) that it was not correct, (iii) that it was correct in one sense and wrong in another, and finally (iv) there was the option of answering ‘cannot say’. Predictions for internalism and externalism are the same as before, but this time the hybrid view entails that the participants answer by choosing the explicitly ambiguous option (‘on the one hand…’). The ‘cannot say’-alternative is introduced to minimise the proportion of guesses or random answers.

The results were the following: 48 % of the answers were externalistic, 22 % internalistic, 17 % ambiguous, and 12 % ‘cannot say’. That is, there were over two times more externalistic answers than any other type of answers, just like in the first experiment. Though there were again a substantial proportion of non-externalistic answers, we conclude that the results lend most support to externalism, as this time the answering options contradict each others. Moreover, since the hybrid hypothesis did not receive much direct support, it is possible that the internalistic answers in the first experiment did not after all manifest ambiguity, but rather were due to some intervening variables.

We conclude that the results suggest that natural kind terms have externalistic essentialist cores that make them refer in virtue of the unknown deep structure of the actual samples.

The paper can be found in whole here.

Reference:

Braisby, N., Franks. B., Hampton, J. (1996). Essentialism, word use, and concepts. Cognition, 59, 247 – 274.

Comments

[Daniel Cohnitz]

Hi Jussi!
Congratulations to the publication of the paper.

Cheers,
Daniel

[Daniel Cohnitz]

... congratulations also to the other Jussi and Henry, of course.