Objective and epistemic gradability: Is the new angle on the Knobe effect empirically grounded?

ABSTRACT

According to the New Angle, any explanation of the Knobe effect must be gradable and asymmetric. It has been argued that only Hindriks’ approach meets both criteria. First, we argue that Holton’s hypothesis also meets the criteria. Second, we show that the authors are not justified in taking the criteria to be empirically justified. We have failed to replicate the asymmetry result in two experiments. Moreover, gradability can be objective or epistemic. We show that the New Angle presupposes objective gradability. In our experiments, the patterns of responses to questions about epistemic and objective gradability are the same, irrespective of whether the feature is objectively gradable (e.g., blameworthiness) or not (e.g., intentionality). Our results thus question the extent to which the New Angle is empirically grounded. Moreover, they raise doubt whether the answers to questions about epistemic and objective gradability can be taken at face value at all.

Abbreviations: NRH - normative reasons hypothesis; NVH - norm violation hypothesis; DQ - degree question; DAQ - degree of agreement question

KEYWORDS:

• Epistemic/objective gradability
• experimental philosophy
• intentionality
• Knobe effect
• new angle
• normative reasons hypothesis

Acknowledgments

We are deeply indebted to Katarzyna Paprzycka-Hausman and Katarzyna Kuś for encouraging us to write this paper and their ceaseless support in the process of writing. We would also like to thank Adam Wierzbicki and Ignacy Szczeniowski for their valuable comments on the earlier drafts.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. The hypotheses they evaluate include: Holton’s Norm Violation Hypothesis, Knobe’s Moral Valence Hypothesis, Blame Hypothesis, Sripada’s (2010) Deep-Self Hypothesis, and the gradable versions of the last three. The only theory to fulfill both criteria is the gradable version of Knobe’s Moral Valence Hypothesis. They reject it for another reason, namely the fact that it cannot account for cases where the agent’s system of values is the reverse of that of participants (Hindriks et al., 2016, p. 212), as shown in the Nazi Germany study (Knobe, 2007).

2. We thank an anonymous reviewer for pressing this point.

3. “According to Hindriks (2008, 2011, 2014)) NRH, the indifference of the agent plays a central role in the explanation of the Knobe Effect. Due to his indifference, the agent fails to be motivated by an effect that he should care about. In other words, he ignores a normative reason”(p. 215).

4. “It seems plausible to say that, ceteris paribus, the less someone cares about a harmful side effect, the more she will be blamed. It also seems unobjectionable to say that, ceteris paribus, the worse the effect is, the more blameworthy the agent is … Given these two claims, the idea that comes in sight is that the amount of blame people attribute depends on the extent to which they see a discrepancy between how much the agent should care and how much she actually cares” (p. 217).

5. For a similar point, see, for example, Brennan, Eriksson, Goodin, and Southwood (2013, p. 210): “Of course, not all norms are such as to admit degrees of violation. Most are, however. At the very least, the relation between competing norms will usually not be one of lexical ordering but rather something softer, allowing an agent to optimize his norm violations in such a way as to minimize their overall badness from his own perspective.”

6. We would like to thank an anonymous reviewer for this objection.

7. Holton does not address the blame-praise asymmetry.

8. Moreover, inasmuch as duties are constituted by the norms, what has been said in defense of NVH applies also to the omissions account of the Knobe effect (Paprzycka, 2015, 2016).

9. The details of the experiment are discussed in Kuś and Maćkiewicz (2016).

10. Indeed, one may even raise doubt that two different things are measured. It is widely accepted in psychological research that a very high correlation coefficient indicates that not two but one construct is actually measured.

11. We would like to thank an anonymous reviewer for pressing us on this issue.

Additional information

Funding

This work was supported by the Narodowe Centrum Nauki [2013/11/B/HS1/03939].

Notes on contributors

Tomasz Zyglewicz

Tomasz Zyglewicz is a graduate student of philosophy at the Graduate Center, City University of New York. His research interests include philosophy of language, metaphysics, and philosophy of law. He also holds a degree in law from the University of Warsaw.

Bartosz Maćkiewicz

Bartosz Maćkiewicz is a graduate student of philosophy at University of Warsaw. His work focuses on application of methods from corpus linguistics and psycholinguistics to research in experimental philosophy.