12:15-01:45pm: PPM at the Jour fixe
of the Center for Junior Research Fellows
here.
"Pessimistic Meta-Induction and the Exponential Growth of Science", Ludwig Fahrbach (Konstanz)
In my talk I show that the argument of pessimistic
meta-induction (PMI) is inconclusive. The premise of the PMI states that in
the history of science, there have been many radical theory-changes: many
scientific theories were highly successful and therefore accepted as true, but
later rejected as false. (A theory is successful iff it makes correct
observable predictions and fits with the data.) The conclusion of the PMI
states that most currently accepted theories, although empirically successful,
will probably be rejected as false in the future. Thus, the success of a theory is
not a good indicator for its truth.
I claim that the premise of the PMI is not
sufficiently precise. A more precise premise should weight every period of
time by the respective amount of scientific work done in that period of
time. Most of the scientific work ever done has been done by the last two
generations of scientists (over the last 50 years or so). Hence, the
preponderance of weight should lie on the last 50 years. On the other hand,
almost all examples presented in the literature to support the premise of the
PMI (case studies about the ether, phlogiston, caloric fluid, etc.) concern
very early parts of the history of science, namely the 17th, 18th and 19th
century. These examples cannot support a correctly weighted premise.
When we
examine the last 50 years we observe that the premise of the PMI is wrong;
most scientific disciplines experienced very few revolutionary theory-changes
in that time. Thus, we observe great stability of scientific theories in the
last 50 years which is what we should project into the future.
"Quantum Theory and Causality - Recent Experiments", Brigitte Falkenburg (Dortmund)
Recent "which way"-experiments of quantum optics will be discussed.
The most interesting of them employ quantum erasers and delayed
choice of the decision about the information which is read out.
Which light do different concepts of causality shed on these
experiments, and vice versa?
"Counterfactuals and Entropy", Matthias Frisch (Baltimore)
David Albert and Barry Loewer have argued that there is a certain class
of probabilistic counterfactuals, central to causal reasoning, that
exhibit a temporal asymmetry and that this asymmetry can be accounted
for by appealing to entropy considerations. The aim of this talk is
to critically examine Albert's and Loewer's proposal.
"On the statistical viewpoint on 'entropy increase'
- A Reminder on the Ehrenfests' Urn Model", Domenico Giulini (Freiburg)
In statistical thermodynamics the 2nd law is
properly expressed in terms of conditional
probabilities. As such it states the likely
increase of entropy without (sic) inducing a
time orientation. This statement itself is
not new. It was emphasized by Paul and
Tatiana Ehrenfest and later by C.-F. v. Weizsäcker.
In my talk I give a quantitative description of
this idea in terms of the Ehrenfests' "urn model",
that makes essential use of Bayes' rule.
This may or may not be new and/or interesting.
"Shedding New Light on Zadeh's Criticism of Dempster's Rule of
Combination
", Rolf Haenni (Bern)
Shortly after proposing Dempster's rule of combination as a
general aggregation rule for evidence from independent sources, Zadeh
formulated an annoying example for which Dempster's rule seemed to
produce counter-intuitive results. Since then, many authors have used
Zadeh's example either to criticize Demster-Shafer theory as a whole,
or as a motivation for constructing alternative combination rules. This
paper shows that the counter-intuition of Zadeh's example is not a
problem of Dempster's rule, but a problem of Zadeh's model that does
not correspond to reality. Two different ways to fix the problem will
be discussed, showing that Dempster's rule of combination perfectly
behaves as one would expect.
"Prediction asymmetries of
generalizations and instantiations", Berna Kilinc (Istanbul, Konstanz)
TBA
"TBA", Christian List (London)
TBA
"Utilitarianism and prioritarianism", David McCarthy (Edinburgh, Konstanz)
Broome has argued that a reformulation of Harsanyi's famous
representation theorem should be given a utilitarian interpretation and that
the theorem shows that prioritarianism is more or less meaningless. I prefer a
different reformulation of the theorem and different argument for a
utilitarian interpretation. But the same kind of argument when applied to a
different representation theorem leads to the conclusion that at least one
form of prioritarianism is both meaningful and reasonably
plausible. Rabinowicz has responded to Broome by arguing that a different form
of prioritarianism is meaningful, and that it gives good expression to
Parfit's informal ideas about the priority view. However, I will argue that
that form of prioritarianism is very implausible. The talk will include a
general discussion of the way uncertainty is involved in the way we ought to
understand utilitarianism and prioritarianism.
"Unmasking the doctrinal paradox:
Group decision-making without fear", Gabriella Pigozzi (Konstanz)
TBA
"What's so special about general relativity?", Oliver Pooley (Oxford)
In formulating the general theory of relativity, one of Einstein's
goals was to generalize the restricted relativity principle of
special relativity to a principle that upheld the physical
equivalence of reference frames in arbitrary states of relative
motion. General covariance was supposed to be one way of satisfying
this requirement. However, most now believe that, with appropriate
mathematical reformulation, any theory can be made to satisfy the
principle of general covariance. Many, instead, view the absence of
non-dynamical fields, rather than its general covariance, as
definitive of what's special about general relativity.
This orthodoxy is now being questioned by a small minority,
including Earman, Rovelli, and Stachel. The following are sometimes
held to be the features of general relativity that really
distinguish it from pre-general relativistic theories:
- 1. the group of diffeomorphism of the spacetime manifold is a gauge
group.
- 2. The genuine physical magnitudes of the theory are
diffeomorphism-invariant.
- 3. Its fundamental entities have no individuality that is independent
of the relational structures in which they are embedded.
These features are typically seen as following from (or amounting
to) general relativity's satisfying a substantive principle of
general covariance.
In this talk I side with the orthodox view. Generally covariant
formulations on non-generally relativistic theories satisfies these
features just as much as general relativity does. What's special
about general relativity is the absence of background fields.
Further clarity is gained by distinguishing notions that are often
conflated, namely, those of non-dynamical fields, of absolute
objects, and of background fields.
"Statistical Inference, the Problem of Induction, and the Realism Debate", Jan Willem Romeyn (Groningen)
Statistics plays an important role in how science solves the
problem of induction. In my talk I first make precise what role it plays, and
then investigate the extent to which, in this role, it can support the realist
ambitions of science.
The first task involves a critical analysis of the
logical empiricist views of Carnap, and a reformulation of inductive
inferences as Bayesian logical arguments. The second involves a reversed
application of De Finetti's representation theorem, and a rather delicate mix
of his strict subjectivism with the frequentist theory. However, these reform
measures do not yet go far enough. In the last part of the talk will argue
that scientists have good reasons for employing underdetermined statistical
models, and sketch a theory of Bayesian abduction.
"The Frequency Theory of Probability: A Defense", Gerhard Schurz (Düsseldorf)
(1) A random experiment ist a kind of causal process under
partly fixed initial and boundary conditions which produces some outcome of a
space of possible outcomes. (2) In a deterministic random experiment the
outcome is always the same. I argue that an ordinary empirical disposition
property such as "being soluble" is a property of a (deterministic) random
experiment, saying that every realization of it would lead to a certain
result. Empirical disposition predicates involve counterfactual idealizations
such as "even if the experiment has de facto only be performed in our universe
n times, if it would have been performed more than n times, then the outcome
would still be the same". In general, it can be doubted whether 'empirical'
disposition predicates can really be called "empirical" , because they involve
counterfactual idealizations. But they are not theoretical predicates, because
their meaning is not determined by a specific theory. (3) In a probabilistic
random experiment the outcomes can be different. But their limiting
frequencies converge against a frequency limit which is called their
statistical probability. This is an even stronger idealization, because it
means: "even if the experiment has been performed de facto only n times, if it
would have been performed infinitely many times, then the relative frequency
of the outcome would converge against a certain real number r". I argue that
there is no philosophical difference between empirical disposition predicates
such as "being soluble" and statistical probabilities such as "the probability
of a coin landing head equals 1/2". The concept of a statistical probability
is a highly idealized but still empirical disposition property. It is not a
theoretical property in the narrow sense. (4) However, there is a big
difference in technical complication: we must fix the sequences in which the
frequency limits are determined. This problem, I argue, is solved in von
Mises' theory of admissible place selections. The ground sequence belonging to
a probabilistic random experiment is the sequence of all performances of this
random event, ordered in time, in an idealized infinite continuation of our
universe. This is not a stronger counterfactiual setting than what we already
assume in empirical disposition predicates. Real sequences such as "my coin
tossing" or "your coin tossing" are always obtained by certain place
selections from that ground sequence. In so-called random sequences these
place selections do not depend on the outcomes. These leads to the von Mises -
Church notion of an admissible place selection as a computable function f: N
-> {yes, no} which may depend on previous outcomes e(m) for m
"Simulating Democracy -
Modeling as Theorizing vs Modeling as Problem Solving", Peter Towbin (Santa Cruz)
TBA
"Voting power: towards a revision of the paradigm", Federico Valenciano (Universidad del Pais Vasco)
TBA
"Costs and gains of a Hamiltonian formulation of general relativity", Christian Wüthrich (Pittsburgh)
In order to enable a canonical quantisation of general relativity,
the latter needs to be cast as a Hamiltonian system. But can a theory which is
generally covariant be forced into the Hamiltonian formalism? If so, how is
general covariance encoded in such an approach? In my talk, I will try to offer
answers to these questions and to then assess more generally the costs and
gains of casting general relativity as a Hamiltonian system with constraints.
