How to Idealize

Interview by Richard Marshall

'I do not see my work as arguing against the anti-realist so much. What I have attempted to do really is to show that the realist/anti-realist debate involves something of a false dichotomy.'

'Idealization is about simplifying things whereas approximation is about distance from the actual truth in modal space (that does not necessarily involve simplification).'

' I don’t believe that there are mathematical objects at all myself. I think that number terms are indexicals and do not refer independent of context.'

'The issue is that complete and consistent world descriptions are massively complex and beliefs about them are demonstrably unsafe.'

'Williamson’s E = K thesis entails that no approximation can ever be evidence. This is totally at odds with every day and scientific uses of approximations as evidence.'

'I reject all a priori knowledge and I reject standard forms of conceptual analysis. I think all questions are ultimately empirical and x-phi is a strong step in the right direction as revolutionary and naturalistic revision of philosophical methodology.'

Michael Shaffer is a philosophy professor interested primarily in epistemology, logic and the philosophy of science. here he discusses the realistvs anti-realist debate, the status of idealisations, counterfactuals, the distinction between idealisation and approximation, intentional relational systems, possible worlds as mental modes and some bad implications for Williamson's Evidence = Knowledge thesis. He denies the analytic/synthetic distinction and thinks x-phi on the right road. Plus he completes a philosophical survey.

3:16: What made you become a philosopher?

Michael Shaffer: I became a philosopher primarily due to the influence of my father. He was a physics professor and the chair of a physics department for about four decades. But, he was always deeply interested in philosophy and the humanities as well. Intellectual discussion was the norm in our household, as well as lots of sports, good food and gardening. He was very supportive of my interests, whatever they happened to be, but I began my undergraduate studies in physics. That quickly changed though and I found that I was more interested in philosophy than physics. Initially, I was attracted to the philosophy of science, logic, epistemology and the philosophy of mathematics (all very broadly understood), but I quickly developed additional interests in ethics too (especially as they relate to both pseudoscience, physical medicine and psychiatry). Also, I should acknowledge that my older brother is a well-known physicist too and I suspected that it was better to be the best philosopher in the family than it would be to be the third best physicist.

3:16: One thing you’re interested in is discovering whether scientific theories are true. You defend a realist position and say they are, against the anti-realist who claims that they aren’t. Nancy Cartwright wrote that science lies – is this the position you’re arguing against? What were her reasons and generally, what do anti-realists argue when they dispute realist claims?

MS: I do not see my work as arguing against the anti-realist so much. What I have attempted to do really is to show that the realist/anti-realist debate involves something of a false dichotomy. As I see it, the issue behind the debate is not best understood as a matter of (strict) truth versus falsehood with respect to theories. Rather, I think that the real issue is one of how realistic theories in fact are. That is to say, the issue is one of how comprehensive they are in terms of how they represent the world and how close to actuality are the worlds they describe. Idealized theories are then just true under simplifying counterfactual assumptions whereas the strictly true theory of a given domain holds independent of any such assumptions. Approximately true theories are true of close possible worlds. So the real issues are how idealized a given theory is and how close to the truth it is, not just whether it is simply true or false. In virtue of this view, one can also then hold that the aims of science involve both truth (as per realism) and utility (as per standard forms of anti-realism). These aims need to be balanced, especially when we lack epistemic resources or simply do not need strict truth. I call this view practical realism and it makes the explanation of scientific progress and practice much more realistic.

3:16: You defend a view that agrees that the truth of a scientific theory is only true of the world of the model, which isn’t ours, isn’t the real world, don’t you? Haven’t you already conceded too much – if the realism is only of a model world, one which doesn’t really exist, then surely whatever the model claims to be true isn’t true of reality?

MS: This is not quite correct. When idealizations are regimented as (simplifying) counterfactuals we can see that they are about relationships between the actual world and idealized worlds (i.e. idealized models of the world). Such counterfactuals are assertions that were certain simplifying claims that are false of the actual world true, the actual world would be as the idealized model depicts it.

3:16: Ha, well, having got everything wrong so far let me ask this: What is the metaphysical status of these model possible worlds? Are you committed to one particular metaphysical understanding of them, or do you think your position is compatible with a range of ontologies – and if so, can you sketch what the likely candidates are?

MS: I think that they are most plausibly understood to be mental representations. Such mental models can be formally laid out as sets of sentences or sets of propositions (as they are in terms of linguistic ersatzism or propositionalism about worlds respectively), but not without introducing problems. Specifically, there are all sorts of computational problems that arise with respect to such linguistic/propositional representations of worlds that negatively impacts their comprehensibility and knowability.

3:16: Why is it important to not confuse the concept of approximate truth with idealization? Is this confusion the main reason for anti-realist claims? And how should we understand idealization?

MS: It is important not to conflate them because (at least as I see it) idealization is about simplifying things whereas approximation is about distance from the actual truth in modal space (that does not necessarily involve simplification). The former is about relations between the actual world and simplified worlds, whereas the latter is about relations between the actual world and other complete worlds.

3:16: Does rejecting the complete worlds assumption commit you to a non-classical logic with truth value gaps, rejection of bivalence and the law of excluded middle and so forth? And again, if it does, doesn’t that weaken your realist credentials – wouldn’t that mean you’d be committed to having some scientific claims that are neither true or false? Just as a matter of what gets said, we don’t have scientists claiming they’ve made a great discovery that is neither true nor false – so doesn’t this position fail to track what scientists claim and undermine realism?

MS: I don’t think that it does if you reject gaps in favor of the view that reference failures induce falsehood.

3:16: What is an intentional relational system? And wouldn’t abstract maths objects fit the bill here too, in which case, does your view commit you to realism about maths objects?

MS: An intentional relational system (as understood by Chris Swoyer) is just a world description that has built into it particulars, relations and non-relational properties. Worlds so understood then are not just extensionally characterized and the relations and properties they incorporate are not reducible to any feature(s) of extensional worlds. I’m not sure about purely mathematical worlds, but if one were a realist about numbers it may well be that given such a view there would be Pythagorean possible worlds describable as an intensional relational system. But, I don’t believe that there are mathematical objects at all myself. I think that number terms are indexicals and do not refer independent of context. Where they are used to refer (in applied mathematics), they refer to arbitrary collections of objects.

3:16: So can you summarise the role of counterfactuals in your approach?

MS: I do not believe that such counterfactuals are vacuous and I do not think that the semantics of such claims (necessarily) requires vagueness. They are claims about how the actual world would be, were it simpler in some respect(s).

3:16: You’ve recently argued that possible worlds semantics renders semantic knowledge impossible and you use the preface paradox to show why. Can you sketch for us your thinking here – does it have a bearing on the previous discussion regarding scientific realism? And what issues regarding the safety condition on knowledge does this particular paradox raise and how do you resolve the problem?

MS: This relates to my answer to what I was saying earlier. The issue is that complete and consistent world descriptions are massively complex and beliefs about them are demonstrably unsafe. So, if you back the safety condition on knowledge and the view that semantic knowledge involves beliefs concerning meanings that are equated with sets of possible worlds, then (propositional) semantic knowledge is impossible. I think that the safety requirement on knowledge is unavoidable. So, this is a big part of my motivation for thinking that possible worlds are mental models that do not have the sort of complex structure that large sets of sentences/propositions do.

3:16: You’ve also argued that Timothy Williamson’s commitment to a number of principles regarding assertion and knowledge leads to rejecting a whole class of ordinary propositional statements that are used as evidence. Does this link with your rejection of his arguments for evidence being knowledge – because of your commitments to evidence being only approximately true? Can you say what the problem[s] is[are] here and what alternative norms of assertion do you argue for that avoids the problem?

MS: Yes, that’s it. Approximations and idealizations are ubiquitous in human thinking. Since approximate truths are all false (by definition), Williamson’s E = K thesis entails that no approximation can ever be evidence. This is totally at odds with every day and scientific uses of approximations as evidence. Lots of evidence is approximately true, due to inaccuracies in measurement and the like. Personally, I defend the view that assertion and action are governed by the norm of justified at-least-approximate truth (i.e. approximately true or true). This ties in with my defense of practical realism and my views on evidence as well. Science has as a proximate aim justified at-least-approximate truth, while its distal aim is strict justified truth (for both measurements and theories). I should also add that the safety condition on knowledge (which is central to my current thinking)allows for an epistemic grounding of the medical evidence hierarchy and supports the contention that evidence involves both probabilistic and modal features.

3:16: Is Quine wrong to reject the analytic/synthetic distinction because he wrongly assumes we can make a distinction between empirical and mathematical theories?

MS: There is no analytic/synthetic distinction. About that he is right.

3:16: Do you think that entertaining epistemic paradoxes is not as dangerous to rationality as many philosophers claims?

MS: Yes. Acceptance and belief are not the same thing and no one really fully believes all of the propositions that constitute paradoxes. They are puzzles that often reveal flaws in potentially believable and sometimes interesting claims. Paradox constituting propositions are entertained in terms of acceptance (which does not involve commitment to truth).

3:16: You’re interested in methodologies in philosophy and x-phi is a movement that has challenged several traditional methodological approaches to doing philosophy hasn’t it. What are the traditional approaches it questions, and what alternatives do you think it uses that are most fruitful? Is your own approach x-phi directed in some respects?

MS: I reject all a priori knowledge and I reject standard forms of conceptual analysis. I think all questions are ultimately empirical and x-phi is a strong step in the right direction as revolutionary and naturalistic revision of philosophical methodology. Positing theories is not confirming theories. Traditional philosophy tries to pretend positing and confirming are the same thing by sliding intuition into the equation as a source of putative evidence. But philosophical theories so understood are merely speculative takes on what might be true devoid of real evidence. All evidence is empirical and scientific evidence is the most dependable (i.e. safe) form of empirical evidence.

3:16: An optional survey of one word answers (with additional comments only if you want to clarify something) –

MS: a. A priori knowledge: yes or no? No

b. Abstract objects: Platonism or nominalism? Numbers are indexicals and other sorts of abstracta are mental objects.

c. Aesthetic value: objective or subjective? Subjective.

d. Analytic-synthetic distinction: yes or no? No.

e. Epistemic justification: internalism or externalism? Both.

f. External world: idealism, skepticism, or non-skeptical realism? Non-skeptical realism.

g. Free will: compatibilism, libertarianism, or no free will? No free will.

h. God: theism or atheism? Atheism.

i. Knowledge claims: contextualism, relativism, or invariantism? Weak contextualism.

j. Knowledge: empiricism or rationalism? Empiricism.

k. Laws of nature: Humean or non-Humean? Non-Humean.

l. Logic: classical or non-classical? Both.

m. Mental content: internalism or externalism? Both.

n. Meta-ethics: moral realism or moral anti-realism? Realism.

o. Metaphilosophy: naturalism or non-naturalism? Naturalism.

p. Mind: physicalism or non-physicalism? Physicalism.

q. Moral judgment: cognitivism or non-cognitivism? No opinion.

r. Moral motivation: internalism or externalism? No opinion.

s. Newcomb's problem: one box or two boxes? Neither.

t. Normative ethics: deontology, consequentialism, or virtue ethics? Consequentialism.

u.Perceptual experience: disjunctivism, qualia theory, No Opinion.

v. representationalism, or sense-datum theory? Representationalism.

w. Personal identity: biological view, psychological view, or further-fact view? Biological.

x. Politics: communitarianism, egalitarianism, or libertarianism? Communitarianism.

y. Proper names: Fregean or Millian? Frege.

z. Science: scientific realism or scientific anti-realism? Both.

Ai. Teletransporter (new matter): survival or death? No opinion.

Aii. Time: A-theory or B-theory? Both.

Aiii Trolley problem (five straight ahead, one on side track, turn requires switching, what ought one do?): switch or don't switch? Switch.

Aiv. Truth: correspondence, deflationary, or epistemic? Correspondence.

Av. Zombies: inconceivable, conceivable but not metaphysically possible, or metaphysically possible? Logically but not physically possible (there is no MP possibility).

3:16: And finally, for the curious readers here at 3:16, are there five books you can recommend that will take us further into your philosophical world?