Suppose I flip a coin 999 times and it comes up heads.
I then flip it a hundredth time but don’t look at it.
I would be comfortable saying I know without looking that it came up heads. (Sure, there’s a chance I’m wrong. There’s a chance I’m wrong if I look, too. If “knowledge” denotes the state of absolute certainty, we don’t ever know anything. It makes more sense to interpret “knowledge” as denoting greater-than-threshold confidence.)
Would a contemporary empiricist say that I don’t know that, because I didn’t see it? That I know this, but it’s not novel information? That it’s novel information, but I obtain it through sensory experience? (E.g., observing the previous 999 flips) Other?
I think the contemporary empiricist would say that all the information you have about the thousandth flip comes from your past sensory experience—your experience of the previous 999 flips plus other relevant experience (such as, say, experiences that form the basis for your beliefs about the base rate of unfair coins). The extent to which your belief about the thousandth flip justifiably differs from maximum entropy (or zero information) is entirely attributable to your prior experiences.
OK, so an contemporary empiricist doesn’t deny the possibility of inference. Good.
Does a contemporary empiricist deny the possibility of inference engines being constructed in ways that bias them towards certain conclusions? E.g., that two people might be born with their brains wired such that, given the same sensory experiences, one of them infers A and the other infers B? (In both cases, presumably, the information about A or B comes from past sensory experience, it’s just that the process for getting one from the other differs.)
If not, then I no longer have a crisp sense of what contemporary empiricists and rationalists actually disagree on.
Like I said in another comment, I identify as a rationalist because empiricism, construed literally, does not allow for informative priors, which makes learning impossible. I’m pretty sure, though, that if you brought this up to a philosopher who identifies as an empiricist, the response would be “Well of course that’s not what I mean by empiricism. Informative priors are fine.” But then, like you, I’m not so sure how to interpret the rationalism/empiricism distinction.
Given your definitions of ‘rationalism’ and ‘empiricism,’ an empiricist would need to assert that informative priors, if they exist, either are not “novel information about the world” or are novel information that we derive from experience. We aren’t perfect Bayesian reasoners, and you haven’t defined ‘information,’ so this doesn’t seem perfectly open-and-cut to me.
One approach an empiricist could take would be to deny that our primordial priors (i.e., our earliest expectations), in themselves, constitute information about the world; perhaps we can use them as a handy framework for genuinely informative research, but the framework itself is not knowledge,
Another approach would be to deny that we have expectations before possessing any sense-perception; perhaps neurological development relies extensively upon sensory input from our environments before anything as cognitively complex as ‘expectation’ or ‘belief’ enters the picture.
Suppose I flip a coin 999 times and it comes up heads. I then flip it a hundredth time but don’t look at it.
I would be comfortable saying I know without looking that it came up heads. (Sure, there’s a chance I’m wrong. There’s a chance I’m wrong if I look, too. If “knowledge” denotes the state of absolute certainty, we don’t ever know anything. It makes more sense to interpret “knowledge” as denoting greater-than-threshold confidence.)
Would a contemporary empiricist say that I don’t know that, because I didn’t see it?
That I know this, but it’s not novel information?
That it’s novel information, but I obtain it through sensory experience? (E.g., observing the previous 999 flips)
Other?
I think the contemporary empiricist would say that all the information you have about the thousandth flip comes from your past sensory experience—your experience of the previous 999 flips plus other relevant experience (such as, say, experiences that form the basis for your beliefs about the base rate of unfair coins). The extent to which your belief about the thousandth flip justifiably differs from maximum entropy (or zero information) is entirely attributable to your prior experiences.
OK, so an contemporary empiricist doesn’t deny the possibility of inference. Good.
Does a contemporary empiricist deny the possibility of inference engines being constructed in ways that bias them towards certain conclusions? E.g., that two people might be born with their brains wired such that, given the same sensory experiences, one of them infers A and the other infers B? (In both cases, presumably, the information about A or B comes from past sensory experience, it’s just that the process for getting one from the other differs.)
If not, then I no longer have a crisp sense of what contemporary empiricists and rationalists actually disagree on.
Like I said in another comment, I identify as a rationalist because empiricism, construed literally, does not allow for informative priors, which makes learning impossible. I’m pretty sure, though, that if you brought this up to a philosopher who identifies as an empiricist, the response would be “Well of course that’s not what I mean by empiricism. Informative priors are fine.” But then, like you, I’m not so sure how to interpret the rationalism/empiricism distinction.
Given your definitions of ‘rationalism’ and ‘empiricism,’ an empiricist would need to assert that informative priors, if they exist, either are not “novel information about the world” or are novel information that we derive from experience. We aren’t perfect Bayesian reasoners, and you haven’t defined ‘information,’ so this doesn’t seem perfectly open-and-cut to me.
One approach an empiricist could take would be to deny that our primordial priors (i.e., our earliest expectations), in themselves, constitute information about the world; perhaps we can use them as a handy framework for genuinely informative research, but the framework itself is not knowledge,
Another approach would be to deny that we have expectations before possessing any sense-perception; perhaps neurological development relies extensively upon sensory input from our environments before anything as cognitively complex as ‘expectation’ or ‘belief’ enters the picture.
Or one could adopt a mixed strategy.