For this Monday we are examining general discussions of realism. This will set us up for a more pointed inquiry on Wednesday into the nature of the realist position for which Haslanger argues in her second chapter.
Preliminary Point: The state of the field.
Talk of “realism vs. antirealism” is a bit misleading. It tends to give the impression that there are two polemically divided camps of metaphysicians with the very nature of reality contested between them.
On the contrary, as both Miller (author of the SEP article) and Brock and Mares would like to suggest, questions of realism are generally localized to a particular domain and the various positions taken by realists and antirealists definitely do not constitute a clear binary.
Miller defines Generic Realism (GR) as follows:
Where a, b, c are distinctive objects of a subject matter,
A, b, and c and so one exist, and the fact that they exist and have properties such a s F-ness, G-ness and H-ness is… independent of anyone’s beliefs, linguistic practices, conceptual schemes, and so on.
Brock and Mares define Realism as follows:
Realism about a particular domain is the conjunction of the following two theses:
- There are facts or entities distinctive of that domain
- Their existence and nature is in some important sense objective and mind-independent.
Brock and Mares point out that realism concerns both facts and entities and that it is possible for one to be a realist about facts in a domain, while denying that there are any real entities referenced by those facts. Miller acknowledges this (halfway through 1. Preliminaries), but does not admit such as a “realist” position. Given Brock and Mares discussion on page 2, do you think Miller’s dismissal of facts can be defended?
Broadly defined, an antirealist position denies either the existence thesis or the independence thesis. Importantly, anti-realists cannot coherently reject both theses, as the mind-dependence of any x necessitates the existence of x.
THE EXISTENCE THESIS:
The Quinean argument for realism about existence:
- We should believe that Fs exist if our accepted theories are ontologically committed to Fs.
- The theories we accept are ontologically committed to Fs.
C Therefore, we should believe that Fs exist.
However, such an argument for realism commits us to an ontology that is, in many cases, undesirable.
Quine suggests that we should either
- Eliminate the aspects of our theory that entail the problematic entities
- Paraphrase away such ontological commitments (though this sounds somewhat similar to later discussion of non-factualism, Quines intent here is that we should clarify our language, not treat it as non-referential).
- In the last resort, accept the ontological commitments wholeheartedly (at least insofar as they are extensions of our best current theories).
Brock and Mares go on to discuss non-classical logics as presenting further opportunities for ontological disavowal. These are a bit beside the point; however I was interested in their discussion of a “substitutional” existential qualifier (pgs. 14-15).
It seems to me that to claim that “∃x(x is a magic dragon) just is the sentence “Puff is a magic dragon””is to invoke the earlier distinction raised between realism about facts and entities. Am I wrong?
AGAINST THE EXISTENCE THESIS:
Brock and Mares lay out the antirealist position in a different way. To be an antirealist about Fs is either to:
- Show that our accepted theories are not ontologically committed to Fs.
- Or, that even if our accepted theories are ontologically committed to Fs, there is no reason to believe that Fs exist.
The antirealist strategy:
- If Fs exists, then they must be either sui generis properties or reducible to some other kind of property.
- If Fs are claimed to be sui generis properties, then use Parsimony.
- If Fs are claimed to be reducible, then use Differential Attitudes to show that they are not reducible, and follow through with a knockout Parsimony.
Parsimony follows from the principle of Ockham’s Razor. Brock and Mares argue that, properly understood, the principle of Parsimony asserts that, ceteris paribus, the better theory is the one which posits qualitatively fewer entities.
Thus, superfluous kinds of entities should not be included in our theories/explanations.
Two notes on Parsimony:
Qualitative parsimony counsels minimizing the number of distinct kinds of entities: a theory which posits apples and elephants is worse (ceteris paribus) than one that posits only elephants.
Quantitative parsimony counsels minimizing the number of entities in general: a theory positing 13 apples is worse than that which posits 1 apple.
Qualitative parsimony is the one they take to be the more important form.
Further, Parsimony is to be understood as not merely a deletion rule (counseling agnosticism about the superfluous entities) but a replacement rule (counseling the replacement of belief in x with the belief in ~x). The important point here is that a mere deletion rule does not commit us to antirealism, just not realism. Quick question: why would sustained agnosticism about the reality of the world not classify one as an antirealist? At the very least it seems that you’d have to be wielding a fairly unnatural (or non-lay person) epistemology and resulting ontology. Relatedly, but distinct, is agnosticism a truly viable option in theory construction and mediation?
The antirealist argument from Parsimony is as follows:
- Fs are superfluous to our explanatory needs.
- If Fs are superfluous to our explanatory needs, then Fs do not exist.
C Therefore, Fs do not exist.
It seems that premise 2 overstates the earlier intuition: that “superfluous entities should not be included in our theories/explanations” does not entail their non-existence.
However, 2 may be supported as follows:
We are entitled to the conclusions of inductive inference, especially when we are in command of the total relevant body of evidence.
An argument against the existence of an entity may be structured as such:
- The total relevant evidence to x is represented by our observations of x.
- Nothing observed thus far is an x.
C Nothing is an x.
In other words: lack of evidence in support of x’s existence counts as evidence for its nonexistence.
This just is the principal of parsimony: entities which are not explanatorily necessary for a theory are superfluous and should be eliminated. Here, the evidence or lack thereof for the existence of an entity is that entity’s explanatory power. If an entity has no explanatory power, and thus there is no evidence for it, then it is not real.
However, it seems to me that there are ways to push back on this argument (although admittedly they involve denying some degree of power to induction).
First, the leap from lack of evidence for x to evidence-against the existence of x seems unsafe. Given TRE, this leap becomes safer, yet I still wonder why agnosticism about the existence of x is not the more appropriate position (now I’ve contradicted my early remarks).
Further, given what I believe to be the impossibility of ever holding TRE, isn’t agnosticism even more appropriate? Do our principles of induction really work as confidently as Brock and Mares would like to suggest?
Regardless, if Parsimony holds, then the antirealist about x has a powerful weapon against the realist about x who posits x as Sui generis.
If instead the realist claims that x is not of kind to itself, but is rather reducible to another real entity y, then the antirealist may employ the Differential Attitudes to attack x.
A quick revision to Parsimony: because reducible entities are not truly of a different kind than their reduction, they don’t add to the qualitative milieu of entities which Parsimony counsels against. Thus, Parsimony* is as follows:
Superfluous and irreducibly distinct kinds of entities should not be included in our theories.
As such, the realist is allowed to keep ontological commitments to those kinds of entities which are superfluous only in that they are reducible.
Thus, if x is reducible to a distinct and non-superfluous entity, Parsimony will not work against it.
The anti-realist about x, following the strategy above, will instead attempt to claim that x is not truly reducible and is thus irreducibly superfluous.
This is commonly done by way of Differential Attitudes (DA)
DA asserts that, as regards x, we have attitudes towards x which are dissimilar from our attitudes towards the reduced phenomena y.
The antirealist argument from DA is as follows:
- Subjects have a certain attitude towards x.
- Subjects do not have that same attitude towards y.
C X is not reducible to y.
Parsimony is then employed:
- x is irreducibly distinct.
- X is superfluous to our explanatory needs.
- If x is superfluous to our explanatory needs and x is irreducibly distinct, then x does not exist.
C x does not exist.
I wonder whether this is totally watertight. It seems that the transition from “x is not reducible to y” to “x is irreducibly distinct” is not valid. Wouldn’t all possible reductions of x have to be negated?
ANTIREALIST POSITIONS DENYING EXISTENCE
Error Theory, see below. Generally: In domain y, statements about that domain are false, though we may believe them. “There is a gap between what we actually believe and what we should believe.” Challenges P1 of the Quinean argument above.
Prefix Fictionalism. Claims about domain y are true, but they contain an implicit prefix “according to theory z”. Thus, facts don’t really reference an underlying ontology. Theories are self-consistent fictions, and they are useful. Challenges P2 of the Quinean argument.
Instrumentalism/Fictionalism: claims about domain y are uniformly false, but this is not a problem. Our goals of discourse in domain y are not truth, but pragmatism, or some other value. Denies P1 of the Quinean argument.
Non-Factualism (expressivism): statements about a domain do not represent facts, but rather express some desire or feeling. Denies P2 of the Quinean argument.
MILLER ON THE EXISTENCE AXIS
Like Brock and Mares, Miller describes two ways in which the antirealist may reject the existence thesis in a given domain.
- The distinctive objects of the domain do not exist.
- The distinctive objects of the domain exist, but do not instantiate any of the properties distinctive of the domain.
Miller uses error theory and arithmetic to illustrate the first, and error theory and morality to illustrate the second.
Error theory is the idea that our theories about a given domain are simply wrong, when we make claims about that domain, those claims are false.
Platonic realism about numbers is the idea that numbers exist in some robustly independent sense (not causally related to human minds).
The argument against Platonic realism is pretty cool, so I’ll throw that in here:
- Platonic arithmetic realism is true (posited for reductio ad absurdum).
- If Platonic arithmetic realism is true, then mathematical objects are acuasal and the objects and the facts about them are mind-independent.
- If a causal explanation of arithmetic reliability is possible, then mathematical objects are causal.
- If a non-causal explanation of arithmetic reliability is possible, then mathematical objects and their respective fact are mind-dependent.
- If an explanation of arithmetic reliability is possible, then it must be either causal or acausal.
C1 If Platonic arithmetic realism is true, then an explanation of arithmetic reliability is impossible.
- If it is impossible to explain the reliability of our beliefs about a domain, then we should not believe in the reality of that domain.
C2 Platonic arithmetic realism is true, but we should not believe in the reality (strong sense) of arithmetic.
C3 Platonic arithmetic realism is not true.
I wonder whether C2 is truly absurd. Perhaps there are good reasons why we should not believe in the reality of something, even if it is real?
Miller addresses the second antirealist possibility by the example of Mackie’s moral error theory.
Suffice it to say that Mackie argues as such:
- If there exist moral facts, then they are objectively true and categorically prescriptive facts.
- There are no such facts.
C Moral theories (and the facts that would follow) are not true.
Mackie’s argument is motivated by a reflection on the nature of an objective and prescriptive fact. If there were objective and prescriptive moral facts, they would be of a sort totally alien to our normal perceptions of morality. Mackie seems to suggest that in our perception of action, there is often moral ambiguity. If it were the case that moral facts existed, then they would be imminent in our experience of reality and they would prescribe courses of action (think roadsigns).
I agree with Wright’s response to Mackie’s error-theory. If Mackie is (and he is) committed to some normative understanding of morality, then why not cash out moral truth in accordance with those norms. In other words, holding moral truth to such fake heights seems disingenuous if one is also willing to accept that moral discourse is better if it leads to “social co-operation” and worse if it does not. Can’t the truth of moral facts be determined by their consequences? Would that no longer be a realist position on morality?
THE INDEPENDENCE AXIS
As mentioned above, antirealists may choose to instead attack the independence thesis of Realism.
Brock and Mares lay out two popular ways in which this has been done: Social Constructivism and Response Dependence.
The distinction between causal dependence and metaphysical dependence:
Causal dependence: A domain of Fs causally depends on us if and only if we play an essential causal role in bringing the Fs into existence; that is, the Fs would not have come into existence in the first place had human beings, and our concomitant actions, intentions and mental states not existed.
Metaphysical dependence: A domain of Fs metaphysically depends on us if and only if the continued existence of our minds is required for the continued existence of the Fs.
Causal dependence is trivial and uninteresting to an antirealist. Metaphysical dependence is the crucial concept.
Brock and Mares cash metaphysical dependence out in terms of psychological dependence:
X is psychologically dependent if its continued existence is dependent on the existence of our minds and on the continued intention of our minds.
Thus, money is psychologically constructed and thus dependent on our minds because, were everyone in the world to forget about the concept of currency, money would cease to exist (though plenty of printed paper slips would still be running around).
This account of social construction seems quite similar to Sally Haslanger’s. For Haslanger, antirealism is unwarranted in the domain of gender dynamics because the power structures that disadvantage women are not merely metaphysically dependent on our minds, but are rather embodied by all people and materialized in our social world, to the point where they become “natural” and only causally (trivially) dependent. Is this compatible with her discussion of objectivity and objectification?
RD is the idea that some facts about entities are mind-dependent in that their a priori truth conditions implicate our mental responses to the respective entities. For example: x is humorous if x arouses laughter.
RD: A concept F is response-dependent if and only if there is a bi-conditional of the form “x is F if and only if x is disposed to produce response R in subjects S in conditions C” that is knowable a priori.
Brock and Mares make four important points of clarification:
- It must be a priori. The authors suggest that affirmations of the mass of objects are somehow “too fortuitous” to count. I need further clarification on this first point, is it that this is somehow a posterior?
- The subject of the bi-conditional must be a mind.
- Germane responses only include changes of mental states.
- The C conditions must not be stated in a trivial “whatever it takes” way. In other words (I think) you can’t try too hard to make it seem a priori response-dependent.
Interestingly, response-dependence does not entail metaphysical dependence. Because the properties of objects which entail responses are intrinsic to those objects, it would not matter if humans ever existed, those properties would still exist.
Given that RD advocates intrinsic natures, can it really be a principle of antirealism?
Brock and Mares argue yes:
- Response-dependent concepts are essentially connected to mental states. RD Properties of objects exist only insofar as they are capable of producing sensations in us, given the proper back ground conditions. (Which, to address the previous worry, can include human existence).
- There is no reason to think that even any mind will experience the same properties of the object, only minds like ours.
- Because RD concepts can be known a priori to be produced within us given the proper C conditions, response dependence is infallibilist; a priori implies necessary. Infallibility is inimical to realism.
What if the realist responds to this argument by becoming a reductionist about the mind? On this account, the mind is real (though wholly reducible to brain chemistry) and red things are red because they produce the “red” response in our minds. However, it seems that by reducing the mental to the physical, the realist would have sidestepped the antirealist motivation for the notion of response dependence. Surely, the antirealist wishes to suggest that there is something wholly fabricated, mental or even contingent (?) about our experience of some concepts. But if our experience of those concepts is determined merely by our brain chemistry, then in what way are those concepts not real? In other words, if humor is just the proper stimulation of certain neurons, how is that any less real than any other causal relationship?