Deceptive worlds, skepticism, and axiarchism

ABSTRACT

Axiarchism holds that fundamental concrete reality is necessarily ordered toward goodness. I argue that it is not fully rational to reject axiarchism while also rejecting radical skepticism. A key premise in the argument is that among conceivable worlds that contain one’s internal duplicate, ‘epistemically inhospitable’ worlds (i.e. worlds where all or most of one’s internal duplicates are radically deceived) are predominant. This predominance of inhospitable worlds provides a prima facie reason for thinking that the actual world is probably inhospitable. To avoid skepticism, this prima facie support for inhospitableness must be countered by a good reason to think that the actual world is probably epistemically hospitable. I argue that opponents of axiarchism lack any such reason. I consider various non-axiarchic ways of dismissing the inhospitable world hypothesis, including appeals to simplicity considerations and to a certain ‘representationalist’ theory of phenomenal consciousness, and find them wanting.

KEYWORDS:

• Axiarchism
• skepticism
• representationalism

Acknowledgements

Versions of this paper were presented at Waseda University, Yale University, The College of William and Mary, and to an online audience from Princeton University and Rutgers University. I’m grateful to audience members for their feedback and questions. Individuals who provided comments or helped in other ways include Lara Buchak, Michael Della Rocca, Keith DeRose, Nathan Dowell, Dan Greco, Verity Harte, Shelly Kagan, Mark Maxwell, Thaddeus Metz, Yujin Nagasawa, Daniel Rubio, Steve Wykstra, an anonymous referee, and the editors of this journal.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 For some recent discussions of axiarchism, see Leslie (2013), Roberts (2014), and Mulgan (2017).

2 See especially Republic VII, 517b.

3 Parfit (1998) discusses the possibility that there could be a probabilistic ‘selector’ that makes certain sorts of worlds more probable. Pruss (2017, 272) also briefly considers the possibility of a ‘stochastic axiarchism’.

4 The definition oversimplifies in assuming that a recent internal duplicate has only one interval of experience that mirrors your recent experience. Some worlds that violate this assumption present special difficulties for a definition of epistemic inhospitableness. Fortunately, my argument does not hinge on how we categorize these problematic cases.

5 The negative argument somewhat resembles Bostrom’s (2003) argument that, conditional on certain assumptions (about the potential for computing power, the proportion of ‘human-level technological civilizations that survive to reach a posthuman stage’ (248), and the interests of advanced ‘posthumans’), it is probable that you are a ‘Sim’, a being living in a computer-based ‘ancestor simulation’ run by posthumans. Whereas I point to the predominance of inhospitable worlds as a (defeasible) reason to think we are in one, Bostrom points to the probability that Sims are predominant among observers as a reason to think you are one. As Bostrom notes (253–254), the Sim hypothesis is analogous to some religious views, as it implies that our (immediate) reality is designed and controlled by posthumans who presumably have godlike powers within our simulation world. Despite points of contact between Bostrom’s argument and mine, there are also philosophically substantive differences. First, the Sim hypothesis is not construed as a skeptical hypothesis (though Bostrom does consider how it might be pushed in a skeptical direction). Second, the reasons favoring the Sim hypothesis are empirical (and thus empirically defeasible), whereas the negative argument supports radical skepticism on a priori grounds. Third, while (non-fundamental) teleology follows as an implication of Bostrom’s radical Sim hypothesis, the negative argument motivates axiarchism (and thus fundamental teleology) as the best grounds for rejecting a radical skeptical hypothesis. (The teleology of the Sim hypothesis does not assuage skeptical worries posed by predominance, since the empirical basis for the Sim hypothesis presupposes that our perceptual faculties are reliable.) Thanks to the editors of this journal for suggesting a comparison with Bostrom.

6 Scenarios like those listed here are considered by Lewis (1986, 116) and Schwitzgebel (2019). The latter thinks possibilities such as these ought to prompt ‘non-trivial doubts about our stability and future’ (23).

7 Button (2013, chap. 15), however, is skeptical about such judgments given the vagueness of the boundary between putative skeptical scenarios that are susceptible to a Putnam-style treatment and those that are not.

8 Chalmers (2010, 489) similarly maintains that mental terms are among the ‘semantically neutral’ terms that refer to the same thing when used by us and our BIV twins.

9 Rinard (2017, 213) motivates external world skepticism in a similar way, noting that the hypothesis that one’s perceptual experiences are veridical ‘imposes strong constraints on the world’, unlike the opposing hypothesis.

10 Lewis (1986, 118), in addressing skeptical worries for modal realism, holds that the cardinality of possible worlds that are ‘deceptive’ is the same as the cardinality of ‘non-deceptive’ possible worlds.

11 Lewis is talking about possible worlds. But the point equally applies to conceivable worlds.

12 Lewis’s sources for this view are Bigelow (1976) and Bennett (1981, 57–61).

13 A defense of axiarchism’s anti-skeptical payoff would have to address challenges to the common view that total value is enhanced by the existence of thinking beings who can perceive and understand the world around them. At present, however, my aim is to raise a skeptical worry for the opponent of axiarchism rather than to show that axiarchism provides a fully satisfactory response.

14 See, for example, Summa Theologica I:4:2 (i.e., question 4, article 2 of the first part).

15 In Summa Theologica I:5:2, Aquinas argues that while goodness and being are convertible, being is prior in thought to goodness, implying that being is explanatorily more basic.

16 See Dretske (1995) and Tye (1995) for defenses of wide representationalism. For defenses of narrow representationalism, see Chalmers (2004) and Seager and Bourget (2017).

17 While to my knowledge wide representationalists haven’t discussed Boltzmann Brains, they have discussed ‘swampman’ (Davidson 1987, 443), a perfect duplicate of a normal human being who arises fortuitously from a swamp struck by lightning. Dretske (1995, 144–151) acknowledges that his wide representationalist theory implies that swampman would not be conscious. Tye (1995, 153–155) is a wide representationalist who holds that swampman would be conscious.

18 Miracchi (2017) argues that the best counter to radical skepticism is to appeal to a view she calls ‘perspectival externalism’, which says that external facts about ‘a subject’s reliable relations to her environment not only play a crucial role in determining the contents of her mental states and events, but also what it is like for her to grasp those contents’ (371). While compatible with wide representationalism, Miracchi’s perspectival externalism is weaker since it does not rule out the possibility that there are some non-representational phenomenal properties that do not depend on external facts. (So a Boltzmann Brain might have some conscious states, just not phenomenal states that purport to represent the world as being in such and such a way.) The objections I raise against an appeal to wide representationalism would equally apply to Miracchi’s appeal to perspectival externalism.

19 Like Putnam’s (1981) anti-skeptical argument discussed in Section 2, the anti-skeptical argument from wide representationalism relies on the fact that the history of external causal influence on a subject helps determine the representational content of the subject’s phenomenal states. But as a theory of consciousness and not merely a theory of semantic content, wide representationalism is stronger in its commitments and in its anti-skeptical potential (Miracchi 2017). Someone could agree that reference is constrained by historical causal influences (if there are any) while also holding that a newly constituted Boltzmann Brain that is a physical duplicate of my brain would have phenomenal states internally like my own. Putnam’s philosophy of language considerations do not clearly rule out the possibility that such a Boltzmann Brain would have a radically mistaken outlook (perhaps due to mistaken beliefs about other minds or about the adequacy of its concepts to describe fundamental reality, as mentioned in Section 2, or perhaps because its ‘external world’ concepts suffer from systematic reference failure).

20 For a helpful overview of objections, including the charge the charge that wide representationalism entails something like epiphenomenalism, see (Bourget and Mendelovici 2014).

21 Though Pruss (2017, 273) contests the view that simplicity favors the standard Big Bang Hypothesis over the five-minutes-old universe hypothesis.

22 Though ‘anti-natalist’ arguments like those of Benatar (2006) challenge the common view that a world oriented toward goodness would include living beings. I pass over such arguments here, since my aim is not to defend axiarchism but to challenge the non-skeptic who denies it.

23 See (Leibniz 1969, 487). The quotation is from Leibniz’s 1697 essay, ‘On the Radical Origination of Things’.

24 For a defense of a priori justification to reject such skeptical hypotheses, see (White 2006; Cohen 2010; Wedgwood 2013; DeRose 2017, chap. 7).

25 See especially (Pryor 2000).

26 See Bostrom (2002, 162) for discussion relevant to these different ways of modeling the selection procedure by which we come by our experiential evidence.

27 Kotzen (2020) points to experiential coherence as evidence against theories predicting Boltzmann Brain superabundance.

28 In support of this claim, consider the following recipe for characterizing inhospitable worlds where experience is typically coherent. First, start with some world w where (i) you have no recent internal duplicate and (ii) if there are episodes of phenomenal experience, they are typically coherent. Next, consider world w* that is just like w except that it contains a single additional substance: a causally isolated, unembodied and conscious monad that exists for a short period of time and whose entire stream of experience is phenomenally just like your recent experience. Any such w* will be an inhospitable world where experience is typically coherent. It seems clear that the set of all such w* worlds exhibits greater internal variety than the set of hospitable worlds. And the w* set does not come close to exhausting the ways that worlds could be inhospitable for you despite the fact that experience is typically coherent.

29 Kriegel (2002) argues that Tye’s appeal to ‘poise’ is inconsistent with the aim of reducing phenomenal facts to representational facts. See Morgan (2018) for related criticisms of other representationalist accounts.

30 I am not aware of anyone else who discusses the possibility that non-occurrent memories might be conscious and the skeptical worry this poses. But see Roelofs (2019) for extended discussion of the possibility that your mind is composed of smaller minds that are also conscious.

31 My discussion here takes some inspiration from an example discussed by DeRose (2017, 246–247).

32 Thanks to an anonymous referee for suggesting I address this response.

33 For more on the modest ‘indifference’ principle motivating this claim, see (Elga 2004).

34 Someone who withholds judgment on the probability of axiarchism cannot help themselves to an axiarchic reason to think that the world is hospitable; and according to the negative argument, there is no good non-axiarchic reason to think this.