Routledge Contemporary Introductions to Philosopy: Philosophy of Science by Alex Rosenberg ---- Chapter 1: Why philosophy of science? History presents many examples of compartments of philosophy breaking away to become science (geometry, physics, biology, psychology, computer science). As topics shift from philosophy to science, there are still problems about these subjects left to philosophy. Deductively valid argument -- if premises are truth, then the conclusions must be true. Inductive argument -- the premises support the conclusion without guaranteeing its truth (the sun will rise tomorrow). One definition of philosophy: philosophy deals with (1) questions science cannot (yet) answer and (2) questions about why science cannot answer them Why is scientific knowledege priviledged? (Alt: Are we overconfident in science and its methods?) Newtonian mechanics as implying determinism. Darwin's destruction of a teleological world view. Quantum indeterminacy as a threat to agent responsibility. Science as the only unique contribution of European (including Middle Eastern) civilization. ---- Chapter 2: Explanation, causation, and laws How does science differ from other enterprises which seek to satisfy wonder and help us understand the world? What are science's standards of what goos as a good (and better) explanation. Empiricism -- knowledge is justified by experience, knowledge of science made up of contingent (non-necessary) truths. Logical positivism -- aimed to combine empiricism and logic to show problems could be solved through language (first half of 20th century_. Possible criteria for scientific explanation: 1) gives necessary and sufficient conditions; 2) derived from general laws; 3) causal explanation 4) predictive -- good grounds for expectation Causation and Hume. Platonists -- science seeks an account of scientific explanation that makes it an objective relation between facts that we set out to discover. Hempel and the deductive-nomological (DN) model: 1) explanation must be a valid deductive argument; 2) explanans contains (and uses) a general law; 3) explanans are empirically testable; 4) explanans are true. Some possible features of scientific laws: 1) laws are universal statements (all As are Bs); 2) laws hold in counterfactuals; 3) laws are (physically and nomologically) necessary truths. Progress in science is often made in the construction of counterexamples and subsequent revisions. Inductive-statistical model (IS) -- explanation raises the probability of the conclusion. ---- Chapter 3: Scientific explanation and its discontents A scientific explanation is one that 1) reveals causal structure (says Salmon); 2) unifies beliefs (say Friedman & Kitcher). Causation -- Aristotelian efficient cause is an immediately prior event which gives rise to an effect. Problem 1: Distinguish causal seqences from coincidences. Problem 2: Causation must capture necessity and sufficiency. Unification -- Understanding increases as number of independent explananda (axioms) needed decreases. Ceteris paribus clauses -- "other things being equal". Two types of probabilistic laws: 1) those that summarize the state of our partial knowledge; 2) those of quantum physics with unexplainable probabilistic propensities. Probabilistic propensity -- disposition to exhibit a certain behavior with a certain frequency. Scientific explanation may be causal, unificatory, nomological, statistical, deductive, inductive. These explain how, but (perhaps) not why. Teleological explanations -- Aristotelain final causes, explanations of why in terms of design and desire/action, not accepted by many philosophers. Scientific explanation should uncover underlying mechansms responsible or the course of nature which are necessarily true. (Leibniz & Kant believe this is what science does.) Leibniz -- Once physical knowledge is complete, all laws will fit together so tightly that a difference in one will unravel the whole structure. (The belief that such a logical consistency yields explanation may fall to the charge of underdetermination.) Kant -- Scientific laws are not logical (analytic) truths; they are synthetic truths which can explain something about the world. Synthetic truths can be known a priori (with observation, experiment, etc) because they are necessary. Hume -- Problem of induction: If we could be certain of the uniformity of nature, we'd have support for scientific laws. But we cannot know that the future will be like the past via pure reason, nor can we rely upon induction to support induction. This is often taken to claim that science is inevitably fallible and scientific knowledge cannot be justified by experience. Is science discovery or creation? ---- Chapter 4: The structure and metaphysics of scientific theories Science theories go beyond explaining a particular phenomena to explain the explanation (explain the empirical generalization); theories unify and go beyond to find underlying processes to account for regularities. How do the parts of a theory work together to explain a diversity of different phenomena? 1) as an axiomatic system (such as Euclid's geometry) 2) by revealing the causal structure of underlying processes Theories which make claims about the unobservable (Newtonian forces, Mendel's genes, molecules, atoms, quarks). Hypothetico-deductivism -- Frame hypotheses which cannot be tested directly; deduce testable consequences from the hypothesis. Intertheoretical reduction -- The derivation of laws of one theory from the laws of another. Example: Kepler's celestial and Galileo's terrestrial motion unified by Newton's more general theory of which they can be seen as special cases. The power of Newton's unification was further sustained as more phenomena were seen to be explained by it (eclipses, comets, shape of the earth, tides, preccession of equinoxes, etc.) Scientific change can be seen as progress, measured in large part by reduction. Axiomatic systems are attractive as explaining how more general underlying mechanisms can explain less general ones. Simplicity at the bottom leads to complexity at the top. Scientific progress involves correlation and improvement of predictions and explanations. Sometimes theories supersede by replacement rather than reduction (Lavoisier's oxygen replaces phlogiston). Science is usually considered to be empirical and testable, yet it relies on thing which are unobservable/unknowable/"occult" (such as gravity). Why are explanations which invoke gravity, genes, etc, better than those which invoke astrology? The problem of justification can be expressed as about the meaning of words and the learnability of language. Theoretical terms can have empirical content. Unobservable objects manifest themselve in experiments; we can see their effects. Principle of verification -- A test to distinguish legitimate theoretical terms of science from the illegitimate. Scientific realism -- The theoretical commitments of science are real. The best explanation for why science works is that it is (approximately) true. (Note: this is an example of an inference to the best explanation.) Instrumentalism -- Alternative to scientific realism. Scientific theories are useful tools but not literal claims that are true. History of science shows a cyclical pattern of realism and instrumentalism among scientists. (Realism in the 17th and 19th centuries; instrumentalism in the 18th and 20th.) Constructive empiricism -- Less common third alternative to both realism and instrumentalism, developed by Bas van Faassen. We can be agnostic as to whether scientific theories are true as long as they enable us to control and predict phenomena. Science may be more than an instrument, but we can't tell if it's more than an instrument. Models -- Theories rely on models, which are simplifications which are true by definition Semantic account of theories -- Theories are sets of models (formal definitions) and claims about what things in the world satisfy the definitions. Does science show a pattern of explanatory and predictive successes which can only be explained by realism? ---- Chapter 5: The epistemology of scientific theorizing History -- Locke, Berkeley, Hume. Hume's problem of induction -- The validity of the inductive inference cannot be justified without using an inductive argument. Reichenbach's response -- If any method of predicting the future works, induction must work. Scientific laws are fallible because they make claims that go beyond the data; laws hold at all times, but this is obviously not testable. Falsification (Popper, as opposed to confirmation) plays an important role in evaluating theories, but how can a general hypothesis really be falsified? What counts as a positive instance for a hypothesis? Is a green emerald an instance of the hypothesis that all emeralds are grue (Goodman)? Those predicates which are admissible in scientific laws are said to be projectable. Probability -- Hypothesis-testing is a matter of probability and statistics. Confirmation is a matter of degree; a hypothesis or theory becomes more confirmed as more positive evidence is acquired. Bayes -- As we acquire data (e), we can calculate how the data changes the probability of h. Bayes' theorem: P(h/e) = P(h&e)/P(e) = P(e/h)P(h)/P(e). Probability -- There are a number of ways of understanding probability, including as gambers' odds, long-run frequencies, or subjective probabilities. Bayesianism relies on probabilities for each application of Bayes' theorem, as well as for assigning initial prior probabilities. How do observation and evidence enable us to choose between theories? Underdetermination -- At the "end of inquiry" can there be two distinct and equally satisfying theories? Theory seems to be underdetermined by observation alone. Theories are judged according to other criteria besides obsevation, including: simplicity, economy, consistency with other already adopted theories. Are theoretical developments guided by observation and/or non-epistemic facts (fashions)? ---- Chapter 6: The challenge of history and post-positivism What determines the succession of theories which characterizes science's history? Conclusions drawn from Kuhn's Structure of Scientific Revolutions (1962) -- History of science is one of change (paradigm shifts), not progress (theory reduction). There is a trade-off between the old and new paradigm which need not be progress. The succession of paradigms is explained by non-scientific factors rather like changes in fashions. Science has no claims to epistemic superiority. Normal science -- The history of science is one of alternating periods of normal science and revolutionary paradigm shifts. Paradigms drive normal science. Paradigms (and scientists) are not passive; they dictate observations. In periods of normal science there are precision gains, ceritification against previous paradigms, paradigm vindications, and explanations of discrepencies of expectation and observation within the paradigm. Paradigm shifts -- Empiricism seems to fail in providing a precise explication of the role that observation and expirement have in theory certification. Perhaps, then, empiricism isn't the right model. Paradigms (eg., Aristotle, Newton, Einstein, Maxwell, Darwin, Lavoisier) do not triumph through the experimental method. Paradigms replace other paradigms in a revolution. Paradigm shifts follow crisis-situations in which all questions seem to be either already answered or unanswerable within the old paradigm. Incommensurability -- Differing paradigms are literally different world views. New paradigms disagree radically with old paradigms. There is no common vocabulary between paradigms, so paradigms can not be translated into other paradigms. (Newton's theory can't explain Einsteinian mass nor Aristotelian impetus). New paradigms have both explanatory gains and losses, and there is no possibility of reduction between theories. Quine, truth, meaning, and holism -- Necessary truths are true in virtue of meaning; contingent truths are true in virtue of facts about the world. Theories cannot be compared sentence by sentence; they must be compared as a whole, and this is how we can choose between theories. The unit of empirical support is the entire theory. Quine, naturalism and naturalized epistemology -- Philosophy is not prior to science, but continuous with it. Philosophy is rejected as a foundation for science and arbiter of its methods. Science is relevant to solving philosophical problems and advancing philosophical understanding. Physics has special credibility among the sciences. Pragmatism (James, Peirce, Dewey) -- Appeal to practical applied success of science for justification. ---- Chapter 7: The nature of science and the fundamental questions of philosophy Epistemic relativism -- Nothing can be asserted to be true independently of some point of view (paradigm), and disagreements between points of view are irreconcilable. The impossibility of translating key concepts reflects a barrier to reduction among all theories. Feyerabend is a strong supporter or reletavism; Kuhn is often seen as supporting this as well. The history of science might be the outcome of non-cognitive factors (social, political, psychological, economic) rather than truth-tracking. Strong program -- Sociologists of science attempted to explain scientific change without appeal to truth. Scientific agreement is constructed through negotiation of parties with their own agendas. Social critics attempt to "dethrone" science from a position of undue/unjust authority. Social sciences and other doctrines embrace Kuhn and scientific relativism because it frees them from the need to adopt scientific methods to gain acceptance. Science is criticized as having sexist/racist agendas and as providing and defending ways to harm people and their environment. Defenders of objectivity need to take Kuhn seriously and undermine the claims of incommensurability. To show that scientific changes are rational (as opposed to just relative to a point of view), defenders must become historians of science.