Abstract
I argue that, in at least one important sense, the hypothesis that you are a brain in a vat provides better explanations than the explanations provided by standard ways of interpreting our best scientific theories. This puts pressure on anyone who—like me!—wishes to resist taking this radical hypothesis seriously when doing science and scientifically-informed metaphysics. Insofar as our resistance is justified, it can’t be justified simply by claiming that the brain in a vast hypothesis is explanatorily impoverished.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
More carefully, the specific version of BIV that I am interested in is one that is spelled out in such detail that it is empirically adequate: it correctly predicts all of the experiences that we have. Note also that BIV, as I am spelling it out here, is distinct from a hypothesis on which the way the computer stimulates your envatted brain gives rise to a three-dimensional space in which ordinary objects (e.g. your body) exist. (The latter is one way of understanding the proposal in Chalmers 2005.) According to BIV, three-dimensional space and ordinary objects seem to exist, but this is an illusion.
The challenge here is, of course, especially pressing for those philosophers who reply to skeptical arguments by claiming that hypotheses like BIV are explanatorily impoverished. But note that in many cases the kinds of explanatory considerations that these philosophers think tell against BIV are plausibly distinct from the kinds of explanatory considerations that are relevant to my argument. How one defines the notion of an explanatory consideration is somewhat complicated (see the next footnote), but paradigm examples of philosophers who have appealed to something that plausibly counts as an explanatory consideration in response to skeptical consideration include Vogel (1990), Bonjour (2003), and Huemer (2016).
These alternative considerations may even be naturally thought of as explanatory considerations—just different explanatory considerations than the ones I discuss. Note in any case that the distinction between explanatory and non-explanatory considerations is going to be nuanced. To my mind, explanatory considerations are considerations to do with the nature and structure of the explanatory relations that a theory posits and are distinct from considerations to do with the theory’s ontology or ideology, the extent to which the theory is conservative (i.e. coheres with common sense) or fecund (i.e. generates novel, testable predictions), and so on. But many philosophers, especially in the literature on skepticism, have a more broad understanding of explanatory considerations. See, for instance, Beebe (2009), which contains a list of 15 distinct kinds of explanatory consideration, including, e.g. ontological parsimony. I have a preference for keeping the notion of an explanatory consideration more constrained because I worry that otherwise we may end up trivializing inference to the best explanation. If we are willing to consider basically any potentially relevant extra-empirical consideration to be an explanatory consideration, then inference to the best explanation just becomes inference to the best theory. And surely all of us ought to endorse inference to the best theory. In any case, nothing in what follows turns on how one draws the explanatory/non-explanatory consideration distinction.
Readers interested in this sort of argument may also be interested in a series of papers by Marcus Arvan (2013, 2014), which present a distinct but nearby argument. Arvan (2014), in particular, argues for a very specific kind of simulation hypothesis—the P2P Simulation hypothesis—on the grounds that it provides a unified explanation of quantum phenomena including the kind of entanglement phenomena captured in the correlation I discuss in Section 2. The P2P Simulation hypothesis is a very specific version of a radical metaphysical hypothesis according to which the kind of quantum phenomena discussed below arises due to errors in the way that multiple computers running the P2P simulation coordinate with one another. My argument applies to a distinct hypothesis—the BIV hypothesis, and turns on a feature of that hypothesis that is shared by radical metaphysical hypotheses quite widely. Thus the argument below can survive issues that might arise with respect to Arvan’s P2P hypothesis.
Note that the correlation is a correlation in what we see, not necessarily in the way things actually are in the external world. Given that I will claim that the BIV hypothesis—which says that there may well be no particles or magnets or detection screens before us—can explain the correlation, this wording is important.
This is an oversimplification, although the upshot once one takes the details into account is the same. A helpful discussion of the precise relationship with relativity is found in Maudlin (2011).
Bell (1964). A particularly accessible discussion of Bell’s theorem is found in Mermin (1981). More carefully, Bell’s theorem shows that no theory can both explain the EPRB correlation by way of a common cause and match the experimental results in this experiment unless it either allows for a non-local dependence between the two sides of the experiment (see further discussion below) or assumes that there is a massive coincidence between the way experimenters set up the magnets and the way in which the source emits the particles.
Those who are surprised by the idea that any answer to the question “Why does the EPRB correlation occur?” must go beyond the quantum formalism should see the Appendix.
One might claim that only the second of these questions is a request for an explanation at all; the first is really a question about prediction. But I don’t think there’s much to be gained from being so restrictive in our use of the term ‘explanation’.
This is rarely made explicit in the literature, but see the discussion in Fuchs and Schack (2015), where they say that “explanations offered by quantum theory have a similar character to explanations offered by probability theory” and that “probability theory explains the agents’ expectations” (7–8). It is also fairly standard to critique Quantum Bayesians along similar lines to Timpson (2008), who writes that QBism has “troubles with explanation” because “We are not interested in an agents’ expectations that [a certain quantum systems will behave a certain way]; we are interested in why it in fact does so” (600).
On my view, one of the fundamental principles governing scientific (and metaphysical) theory choice is a principle that says that we ought not leave robust patterns like this without (substantive) explanations. The argument for this can be found in Emery (2019) (in general terms) and in Emery forthcoming-b (applied to quantum ontology). But I won’t be taking such a strong stance here.
If one thinks that Humean explanations do not involve identifying the reason why the explanandum occurred, then one will think that those who give a Humean account of the quantum wavefunction (see, e.g. Bhogal and Perry 2017), will also leave the EPRB correlation without a substantive explanation. I will set this position aside, since the claim that Humean explanations are not substantive is highly controversial and there are fairly standard Humean responses (see especially Loewer 2012).
Note that my use of the term ‘non-local’, while a relatively natural one for those familiar with Bell’s Theorem (one of the key assumptions of that Theorem is a denial of non-locality in my sense), is also more expansive than the usage favored by many philosophers of physics. In particular, it subsumes both non-locality and non-separability as distinguished in, e.g., Maudlin (2007). Nothing important hinges on this, however. The main role that the term ‘non-local’ is playing in my argument is in helping to organize the various ways of explaining the EPRB correlation into groups. If the reader prefers a different usage of ‘non-local’, they should feel free to ignore the way in which the various options are organized below and assess my argument regarding each of those options directly.
This is the sort of view put forward in Ismael and Schaffer 2020. As they put it, “… the composite [left and right particle] system is more fundamental than—and in that sense a common ground of—its [left particle] component and its [right particle] component. [The left particle] and [right particle] are derivative aspects (or fragments) abstracted from a more fundamental whole. The nonlocal connection between [left particle] and [right particle] arises from their spatially spread-out common ground.”
Those who are familiar with Maudlin’s (2007) terminology may think of this explanation as involving a non-separable, as opposed to non-local dependence relation. See the footnote above and my definition of non-local dependence relation in the text.
As an anonymous reviewer points out, this claim is highly plausible for dynamical laws, but will be more nuanced insofar as we focus on purported non-dynamical laws like conservation laws. I’m going to set this issue aside, since the existence and nature of the latter sort of law is controversial, and it would take more space than I have available here to do justice to the nuance involved in philosophical accounts of such laws.
This option is salient if one endorses a Bohmian interpretation of the dynamical laws governing quantum systems. If one instead endorses a collapse or many worlds interpretation, there will be just a single field in configuration space.
As is standard (see Ney 2021, p. 89), I am presenting configuration space realism as a view on which three-dimensional space still exists, but just is non-fundamental. Note however that early presentations of configuration space realism sometimes seem to suggest that three-dimensional space does not exist at all. See for instance, Bell (1987, p. 204) or Albert (1996, p. 277). Note that these early versions of configuration space realism are plausibly considered radical metaphysical hypotheses themselves, since they deny the reality of 3D space and 3D objects. In any case, I won’t consider these versions any further here.
For more on how the high-dimensional entities might give rise to three-dimensional entities see Albert (2015), Ney (2020, 2021). Ney (2021) argues that 3D particles are part of the field that exists in high-dimensional space, but notes that the relevant notion of ‘part’ is novel because it must allow for parthood relations between physical entities in distinct spaces. Note that non-compositional grounding relations may be familiar from philosophy (e.g. the relation between a truth and its truthmaker is not plausibly compositional). My point is that they are not familiar in science or the metaphysics of science.
Here all I mean by an inter-spatial dependence relation is just a dependence relation that holds between distinct physical spaces. It is perhaps worth emphasizing here that there are some dependence relations between physical spaces that are relatively familiar. Think, for instance, of the two-dimensional space that includes the face of a cube and the three-dimensional space that includes the cube itself. These two spaces are related in important ways. But this is because the two-dimensional space is a part of the three-dimensional space in a very straightforward sense; the two spaces are not in fact distinct. In the case at hand, the three dimensions of ordinary physical space do not correspond to any three of the dimensions of configuration space.
Note that I am open to the idea (and it is compatible with my argument) that explanatory novelty relative to other standards is also relevant to scientific theory choice; the specific kind of explanatory novelty described above is the main focus of the discussion here, however.
The “qua explanation” in this sentence is important. There is more to a theory than the explanations that it provides. There is the mathematical apparatus it uses for predictions, the underlying ontology it posits, the relations that it stands in with respect to other theories. I have said nothing at all about any of these features of T1 or T2. So, we shouldn’t make any claims as to whether all things considered T2 is more novel than T1. But we can still assert that the explanations that T2 gives of the new phenomena are more novel than the explanations that T1 gives Readers who are concerned that it doesn’t make sense to care about explanatory novelty unless you care about novelty in general should see the discussion in response to objection 2 in Sect. 5.
I discuss some potential reasons for why we should avoid this kind of explanatory novelty in my response to objection 2 in Sect. 5.
Readers who want to agree that there’s something wrong with choosing non-local explanation over the fishtank explanation but don’t think that this has anything to do with explanatory novelty should see the discussion in response to objection 5 in Sect. 5.
The other radical metaphysical hypotheses mentioned at the end of the introduction share the following key feature with the BIV hypothesis: the EPRB correlation is not in fact a correlation between particles that are spatially distant from one another. In the simulation hypothesis it is a correlation in two components of the program that is running the simulation. In the Boltzmann Brain hypothesis it is a correlation in the way that the particles came together to form your brain.
Those familiar with the details of Bell’s Theorem may wonder how the shift to the BIV hypothesis avoids that theorem. There may be more than one way of going here, but perhaps most obviously, given the BIV hypothesis, we no longer need to assume that which properties are assigned to the particles at the source is independent of the setting of the magnets (which is a key assumption of Bell’s proof).
In my favored terminology, as discussed in 3.4, I would say that BIV provides a substantive explanation of the EPRB correlation that, while novel, is not of a novel type. But again, if the reader finds this terminology unhelpful they can run the argument just in terms of degrees of explanatory novelty.
Remember that the EPRB correlation, as I understand it, is a correlation in our experiences. Thus BIV can be empirically adequate. In conversation I sometimes have other philosophers insist that the fact that there is a correlation in the behavior of the particles is a part of the data, and thus that BIV cannot be empirically adequate. But this does little to change the debate, it just shifts the place in which it is happening. If you think that the fact that there are particles and magnets behaving a certain way is strictly speaking part of the data, then you will need to have a relatively complex story about how to determine, based on our experiences, what the data collected in any given case involves. One option for determining what the data is will involve steps analogous to steps 1–3. The point will then be: if that the analog of steps 1–3 is a reasonably good candidate for determining what the data are, then it seems as though we ought to take BIV seriously.
Here is what Schaffer and Ismael say about why: “Why a common cause? It seems wrong to say that [the result on the left] causes [the result on the right], and wrong to say that [the result on the left] causes [the result on the right]. First of all, these events occur simultaneously at arbitrary distances, so if one assumes either that causes must precede their effects (or at least cannot be simultaneous with them) or that causes must operate locally (no “action at a distance”), then one rules out either causing the other. Secondly, causation is generally considered to be an asymmetric relation, yet the relation between [the result on the left] and [the result on the right] seems perfectly symmetric. It would seem arbitrary to position one as cause and the other as effect” (13).
Thanks to an anonymous reviewer for encouraging me to think more about this case.
Perhaps the most common specific version of this objection that I encounter in conversation is that BIV is not taken seriously because it is not testable. But note that BIV is, by stipulation, empirically adequate. This means that there is no experiment that we can perform for which BIV and our best scientific theories will make conflicting predictions, and thus that there is no experiment that we can perform that will allow us to test whether BIV is better or worse than our best scientific theories in terms of correctly predicting the results of experiments. And insofar as this means that BIV is not testable, it also means that our best scientific theories are not testable in precisely the same sense. There is no experiment that we can perform that will allow us to test whether our best scientific theories are better or worse than BIV in terms of correctly predicting the results of experiments. In my experience those who put forward this version of the objection often have something more sophisticated in mind in terms of “testability”, or are motivated by a deeper disagreement about whether BIV can in fact be empirically adequate (see footnote 33). I agree that either of those are interesting strategies to pursue for ultimately ruling out BIV, but I don’t think that they are so straightforward and obvious as to undermine my argument.
See for instance Hesse (1962), especially chapter 8.
Thanks to an anonymous referee for encouraging me to think more about cases like this. Other historical cases that are interesting in this context include Newton’s introduction of action at a distance in his gravitational theory and Einstein’s denial of a preferred reference frame in special relativity.
See, for instance, Huemer (2016).
The latter is the view defended in Emery (2017). A more general defense of the claim that divergence from the manifest image is an important criterion in theory choice can be found in Emery forthcoming-a.
References
Albert, D. Z. (1996). Elementary quantum metaphysics. In: Bohmian mechanics and quantum theory: An appraisal (pp. 277–84). Springer.
Albert, D. Z. (2015). After physics. Harvard University Press.
Allori, V. (2013). Primitive ontology and the structure of fundamental physical theories. In: A. Ney and D. Albert (Eds.), The wave function: Essays on the metaphysics of quantum mechanics (pp. 58–75).
Allori, V., Goldstein, S., Tumulka, R., & Zanghı, N. (2010). Many worlds and Schrodinger’s first quantum theory. British Journal for the Philosophy of Science.
Arvan, M. (2013). A new theory of free will. The Philosophical Forum.
Arvan, M. (2014) A unified explanation of quantum phenomena? The case for the peer-to-peer simulation hypothesis as an interdisciplinary research program. The Philosophical Forum.
Beebe, J. (2009). The abductivist’s reply to skepticism. Philosophy and Phenomenological Research, 79, 605–636.
Bell, J. S. (1964). On the Einstein–Podolsky–Rosen paradox. Physics, 1, 195–200.
Bell, J. S. (1987). Speakable and unspeakable in quantum mechanics: Collected papers on quantum philosophy. Cambridge University Press.
Bhogal, H., & Perry, Z. (2017). What the humean should say about entanglement. Noûs, 51(1), 74–94.
Bohm, D. (1951). Quantum theory. Prentice Hall.
Bonjour, L. (2003). A version of internalist foundationalism. In: L. BonJour, E. Sosa (Eds.), Epistemic justification: internalism vs. externalism, foundations vs. virtues (pp. 3–96). Malden, MA: Blackwell.
Caves, C., Fuchs, C., & Schack, R. (2002). Quantum probabilities as Bayesian probabilities. Physical Review A, 65, 1–6.
Chalmers, D. (2005). The matrix as metaphysics. In C. Grau (Ed.), Philosophers explore the matrix. Oxford: Oxford University Press.
Egg, M., & Esfeld, M. (2014). Non-local common cause explanations for EPR. European Journal for Philosophy of Science, 4(2), 181–196.
Einstein, A., Podolsky, B., & Rosen, N. (1935). Can quantum-mechanical description of physical reality be considered complete? Physical Review, 47(10), 777.
Emery, N. (2017). Against radical quantum ontologies. Philosophy and Phenomenological Research, 95(3), 564–591.
Emery, N. (2019). Laws and their instances. Philosophical Studies, 176(6), 1535–1561.
Emery, N. (forthcoming-a). Mooreanism in metaphysics from Mooreanism in physics. Inquiry.
Emery, N. (forthcoming-b). The governing conception of the wavefunction. In: V. Allori (Ed.), Quantum mechanics and fundamentality.
Foster, J. (2004). The divine lawmaker: Lectures on induction, laws of nature, and the existence of god. Oxford University Press.
Fuchs, C. A., Mermin, N. D., & Schack, R. (2014). An introduction to QBism with an application to the locality of quantum mechanics. American Journal of Physics, 82(8), 749–754.
Fuchs, C., & Schack, R. (2015). QBism and the greeks: Why a quantum state does not represent an element of physical reality. Physica Scripta, 90(1), 015104–4.
Fuchs, C. (2017). On participatory realism. arXiv:1601.04360
Healey, R. (2012). Quantum theory: A pragmatist approach. The British Journal for the Philosophy of Science, 63(4), 729–771.
Healey, R. (2017). Quantum-Bayesian and pragmatist views of quantum theory. The Stanford Encyclopedia of Philosophy, https://plato.stanford.edu/archives/spr2017/entries/quantum-bayesian/.
Hesse, M. (1962). Forces and fields: On the concept of action at a distance in the history of physics. Dover Publications.
Hildebrand, T. & Metcalf, T. (forthcoming). The nomological argument for the existence of god. Noûs.
Hubert, M., & Romano, D. (2018). The wave-function as a multi-field. European Journal for Philosophy of Science, 8(3), 521–537.
Huemer, M. (2016). Serious theories and skeptical theories: Why you are probably not a brain in a vat. Philosophical Studies, 173, 1031–1052.
Ismael, J. (2020). What entanglement might be telling us: Space, quantum mechanics, and Bohm’s fish tank. In D. Glick, G. Darby, & A. Marmodoro (Eds.), The foundation of reality: Space, time, and fundamentality (pp. 139–153). Oxford University Press.
Ismael, J. & Schaffer, J. (2020). Quantum holism: Nonseparability as common ground. Synthese, 197(10), 4131–4160.
Loewer, B. (1996). Humean supervenience. Philosophical Topics, 24(1), 101–127.
Loewer, B. (2012). Two accounts of laws and time. Philosophical Studies, 160(1), 115–137.
Maudlin, T. (2007). The metaphysics within physics. Oxford University Press.
Maudlin, T. (2011). Quantum non-locality and relativity (3rd ed.). Wiley.
Mermin, D. (1981). Quantum mysteries for anyone. Journal of Philosophy, 78(7), 397–408.
Ney, A. (2012). The status of our ordinary three dimensions in a quantum universe 1. Noûs, 46(3), 525–560.
Ney, A. (2020). Finding the world in the wave function: Some strategies for solving the macro-object problem. Synthese, 197(10), 4227–4249.
Ney, A. (2021). The world in the wave function: A metaphysics for quantum physics. Oxford University Press.
North, J. (2013). The structure of a quantum world. In A. Ney & D. Albert (Eds.), The wave function: Essays on the metaphysics of quantum mechanics (pp. 184–202). Oxford University Press.
Timpson, C. G. (2008). Quantum Bayesianism: A study. Studies in History and Philosophy of Science Part B: Studies in History and Philosophy of Modern Physics, 39(3), 579–609.
Vogel, J. (1990). Cartesian skepticism and inference to the best explanation. Journal of Philosophy, 87, 658–666.
Acknowledgments
For helpful feedback, thank you to the ELLMM City group at Yale, the April Fools Metaphysics Workshop, Marcus Arvan, Matt Duncan, Michael Huemer, Ned Markosian, Elizabeth Miller, and Tomoji Shogenji. Thanks also to two anonymous reviewers for this journal, whose comments resulted in significant improvements to the paper.
Funding
There is no funding to report.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix: How our current best science predicts the EPRB correlation
Appendix: How our current best science predicts the EPRB correlation
Those who are unfamiliar with the quantum formalism often have difficulty believing that the formalism itself doesn’t give a clear indication of the correct explanatory strategy. Here is a quick primer on the quantum formalism to help make this point clear.
The scientific theory that predicts the EPRB correlation is the standard non-relativistic quantum formalism. This formalism consists of a set of rules for representing quantum systems using vectors and for generating predictions using those representations. The details are complicated, of course, but here is the only bit that matters for our purposes. We use what is called bra-ket notation to indicate the vectors that represent quantum systems. So, for instance we might use |R> to represent a particle that is in region R and |S> to represent a particle that is in region S. If a system has multiple degrees of freedom, we use tensor products of vectors to represent the quantum state, which we indicate using series of bra-kets. So if we have a two-particle system where particle 1 is in R1 and particle 2 is in S1, we might represent this system using |R1>|S1>. And if we have a two-particle system where particle 1 is in R2 and particle 2 is in S2, we might represent this system using |R2>|S2>.
One of the remarkable facts about quantum systems is that if |X> and |Y> represents possible states of the system and if a and b are complex numbers such that |a|2 + |b|2 = 1, then a|X> + b|Y> also represents a possible state of the system. This means that if we use |R> and |S> to represent possible states of a system then there are some a and b such that a|R> + b|S> also represents a possible state of that system. The quantum formalism is wholly silent about what it means for a system to be in a state represented by a|R> + b|S> except to say that if we perform a measurement, the probability of finding a particle in that state in R is |a|2 and the probability of finding it in S is |b|2.
Similarly, This means that if we use |R1>|S1> and |R2>|S2> to represent possible states of a two-particle system, and if a and b are complex numbers such that |a|2 + |b|2 = 1, then a|R1>|S1> + b|R2>|S2> also represents a possible state of that system. The quantum formalism is wholly silent about what it means for a system to be in a state represented by a|R1>|S1> + b|R2>|S2> except to say that if we perform a measurement the probability of finding particle 1 in R1 and particle 2 in S1 is |a|2 and the probability of finding particle 1 in R2 and particle 2 in S2 is |b|2.
Think back to the EPRB experiment. Call the region at the top of the left detection screen UL, the region at the bottom of the left detection screen DL, the region at the top of the right detection screen UR, and the region at the bottom of the right detection screen DR. The state of the two particles at the end of the experiment can be represented using the following vector:
The quantum formalism says nothing whatsoever about it means for this systems to be in this state except for the following: the probability of seeing a flash at the top on the left and the bottom on the right is 1/2 and the probability of seeing a flash at the bottom on the left and the top on the right is 1/2. It follows that the probability of seeing opposite outcomes is 1 and the probability of seeing the same outcome is 0.
In this way, the quantum formalism predicts the EPRB correlation. But it says nothing at all about what explains that correlation.
Rights and permissions
About this article
Cite this article
Emery, N. Quantum correlations and the explanatory power of radical metaphysical hypotheses. Philos Stud 179, 2391–2414 (2022). https://doi.org/10.1007/s11098-021-01769-z
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11098-021-01769-z