Poker Machine Epistemology

The search for the mythical episteme has a potential new tool: the poker machine. The manufacturers code answers to potential problem situations and assign probabilities of the answers being correct. All the players need do is press buttons to choose the problems then run the bets. Occasionally victory tones will ring along with a shower of tokens on the screen to signify a probable problem solution.

Poker machines can be intoxicating, a good win can deceive us into thinking a pattern of success will follow. One can believe anything without good reason.

We already have ways of guessing solutions to life’s puzzles and problems: dreaming, chatting, looking, listening, smelling, feeling, intuiting, assuming past patterns are portents, ad infinitum. Bayesian inference is one of the commonly preferred poker machines, and there are other brands with labels such as induction and abduction.

In the beginning is the problem, and next off the rank is the guess. Guesses are what we living beings do, mainly unconsciously (e.g. mutations, migration etc), and it is psychologically comforting to think we guess correctly. Indeed through the aeons a lot of the guesses must have been pretty good otherwise there would be no life.

The trouble is, no matter how psychologically tempting it might be to believe in the outcomes of guesses, there is no logical inference that can verify them. No wonder Karl Popper is not popular, for no gambler wants to be told his addiction is based on groundless belief or that his poker machines have limited payouts.

The trouble with gamblers and epistemic philosophers in general is that they give greater dignity to the statement “I know snow is white”, than to the statement “snow is white”, as said Karl Popper on page 141 of “Objective Knowledge” 1972. Why? Are we really seeking truth or not?

We must distinguish between the psychology of science and the logic of science. We must also distinguish between the desire for meaningfulness and objective truth. Falsifiability means we are prepared to put our money where our mouths are…no matter what the outcome.

This entry was posted in Uncategorized. Bookmark the permalink.

4 Responses to Poker Machine Epistemology

  1. Aaron says:

    Re: “… there is no logical inference that can verify them.”

    Popperians cannot avoid making inductive inferences. Since deductive inference is not ampliative, the Popperian cannot deduce from basic observation statements factual information which goes beyond the factual information contained in the basic observation statements themselves, and this entails that scientific theories or hypotheses can say nothing about what is not already observed.

    But as I see matters the problem is worse. Take any general scientific statement, say, “All galaxies have super-massive black holes at their centers.” For a falsificationist this reports little more than that all our best astronomical data have thus far revealed super-massive black holes at the center of galaxies. It says nothing about whether we are likely to observe a black hole at the center of a newly discovered galaxy; and of course this is a desideratum of research programmes: we require theories which they are confident will make certain observations more or less likely.

    At this juncture Popper introduces corroboration as a way to resolve this problem. Popper says that highly falsifiable hypotheses (“bold conjectures”) that pass severe tests are more highly corroborated. In short, Popper uses highly corroborated hypotheses as a means by which to make ampliative inferences – to infer new models or hypotheses, and to select between heretofore competing unfalsified theories. But this strikes me as confirmation, and the inferences from highly corroborated theories to new theories are ex hypothesi not deductive since deductive inferences are not ampliative. So, it must be inductive inference.

    The Popperian does not see through the fog of Hume’s problem of induction. Rather, she attempts to make her way through it on the coattails of others.

  2. Bruce Caithness says:

    You start off saying Popperians cannot avoid making inductive inferences? Why just Popperians, why not anyone? I wonder what you mean by induction – discovering regularities in the world, inventing explanatory theories, “proving” general statements or putting a “p” value on a theory? You might also mean just observing something, like ducks walking across a road, or a binomial curve? If we happen to think there are regularities then we have imposed them on the world, they are pyschologically apriori but not valid apriori. What did Karl Popper mean by the term induction? In “Realism and the Aim of Science” p 147 he stated “By induction I mean an argument which, given some empirical (singular or particular) premises, leads to a universal conclusion, a universal theory, either with logical certainty, or with ‘probability’ (in the sense that this term is used in the calculus of probability).

    You can observe anything, or guess anything, but I believe David Hume pointed out that ampliative conclusions are neither certain nor probable. Observation in itself is not induction. Mind you Popper also made the point that perception is modified anticipation as per the modus tollens:

    If the theory is true the inference is true,
    The inference is not true,
    Therefore the theory is not true.

    Even our perceptual systems are not blank slates. Our continually modified expectations impose regularities onto the world; they are not “induced” into us. The world does not “inform” us. The regularities we impose on the world come from psychological expectations modified by experience. This does not mean they are valid.

    If you can prove that any theory, cosmological or otherwise, is true you are a better man than Sir Isaac Newton, even Immanuel Kant thought his theories were true and unfortunately based quite a bit of his philosophy on it. Einstein made some outlandish guesses didn’t he?

    Corroboration is a report on how well a theory has survived testing, there is nothing ampliative in an historical report. You can guess whatever you like after the testing. Confidence is fine, but it is a guess.

    Reason has a job to do, reasons do not.

    If “critical rationalism” is a perjorative term so be it. It is hard if not impossible to nail the truth about any universal theories but that does not mean we should suspend the effort. It is a pity that the Karl Popper most textbooks discuss bears little relation to the Karl Popper in his own books, a straw man indeed.

  3. Lee Kelly says:

    Popperians cannot avoid making inductive inferences. Since deductive inference is not ampliative, the Popperian cannot deduce from basic observation statements factual information which goes beyond the factual information contained in the basic observation statements themselves, and this entails that scientific theories or hypotheses can say nothing about what is not already observed.

    You are very, very confused. Popperians agree that one ‘cannot deduce from basic observation statements factual information which goes beyond the factual information contained in the basic observation statements themselves’. However, Popperians also contend that one cannot induce ‘from basic observation statements factual information which goes beyond the factual information contained in the basic observation statements themselves’.

    In any case, you then commit a glaring non-sequitur: ‘and this entails that scientific theories or hypotheses can say nothing about what is not already observed.’ What? For this to make any sense, you must be operating with a unstated assumption, but what? Only if we presume that scientific theories must be logically reducible (through some kind of logical inference) to observation, does your conclusion follow. However, this is exactly the assumption which Popperians deny, criticise, and replace. That is, observation is not the source of our scientific knowledge: guesswork and conjecture are the source (sometimes they resemble “induction”, but there is really no such thing). Observation, then, plays a regulative role–it helps select among competing scientific theories through the relation of falsifiability.

    But as I see matters the problem is worse. Take any general scientific statement, say, “All galaxies have super-massive black holes at their centers.” For a falsificationist this reports little more than that all our best astronomical data have thus far revealed super-massive black holes at the center of galaxies. It says nothing about whether we are likely to observe a black hole at the center of a newly discovered galaxy; and of course this is a desideratum of research programmes: we require theories which they are confident will make certain observations more or less likely.

    I think you are supremely ignorant of the philosophy you are ostensibly criticising. The statement ‘All galaxies have super-massive black holes at their centers’ entails for Popperians exactly what it entails for everyone else. It cannot be reduced (inductively or deductively) to any finite set of observations, and it doesn’t have to be. New knowledge cannot be inferred (inductively or deductively) from previous knowledge, whether past theories or observations.

    At this juncture Popper introduces corroboration as a way to resolve this problem.

    This has nothing to do with corroboration, because your critique is just the result of a gross misreading of Popper, and I’m being generous by assuming you read Popper at all.

    Popper says that highly falsifiable hypotheses (“bold conjectures”) that pass severe tests are more highly corroborated. In short, Popper uses highly corroborated hypotheses as a means by which to make ampliative inferences – to infer new models or hypotheses, and to select between heretofore competing unfalsified theories.

    Not at all. Popper explicitly denied that corroboration was ampliative. Indeed, he also denied that probabilistic inference was ampliative. Corroboration was intended as a measure of how well a theory had stood up to tests, and perhaps also a measure of how much our knowledge had grown, (i.e. the logical difference between old and new knowledge).

    In short, corroboration, unlike truth, does not transmit from premises to conclusion in a valid argument. That is, however corroborated a theory, that corroboration (like probabilistic support) does not transmit to through the deducibility relation to any of its consequences. In other words, corroboration is not ampliative–it does not imply that unverified predictions of a theory have any higher probability of being true.

    Much less does corroboration provide a means of inferring scientific theories from observation, since corroboration is defined relative to the two. That is, we must begin with a theory before we can calculate how corroborated it is, so corroboration has nothing to do with inferring such theories.

    The Popperian does not see through the fog of Hume’s problem of induction. Rather, she attempts to make her way through it on the coattails of others.Your pomposity and ignorance are a potent combination.

  4. Lee Kelly says:

    I messed up the formatting on that final quote. Here is the correction:

    The Popperian does not see through the fog of Hume’s problem of induction. Rather, she attempts to make her way through it on the coattails of others.

    Your pomposity and ignorance are a potent combination.

Leave a Reply

Your email address will not be published. Required fields are marked *

please answer (required): * Time limit is exhausted. Please reload the CAPTCHA.