How to Think about Reliability, William Alston
Alston, William, ‘How to Think about Reliability’ in ‘Epistemology: An Anthology,’ eds. Ernest Sosa and Jaegwon Kim, Blackwell Publishers, Massachusetts, 2000
p. 354 - Dealing with the concept of reliable belief formation.
p. 355 - ‘…a particular process that takes place at a particular precise time is not the sort of thing that does or does not enjoy a favourable ratio of true beliefs among its products. It occurs just once; the one belief it produces is either true or false, and there’s an end to it. Hence, as is regularly said by both friend and foe, it is a type of cognitive process, rather than a particular process (a token) that can be assessed for reliability.’
Any given belief might be the result of multiple processes…
p. 357 - ‘Hence a great deal is riding on the supposition that there are no objective, nonepistemic facts that determine a unique type assignment for each token belief formation. It is that supposiiton I will seek to discredit.’
Talks about reliability as a propensity rather than as a frequency.
Claims that when we talk about reliability we talk about it in propensity terms.
p. 358 - Conditional reliability is for belief-dependent processes whilst (straight) reliability is for belief independent processes.
p. 359-60 - Thinks there are ‘natural’ kinds of processes and that this allows us to get around ‘the many causes to one effect’ problem for reliable processes.
p. 360 - ‘If the epistemic status of a belief is a function of the reliability of the process that generates the belief, it is the reliability of the psychological process that is crucial.’
p. 361 - ‘The crucial point is that every belief formation involves the activation of a certain psycholgically realized function. That activation yields a belief with a propositional content that is a certain function (the function the pyschological realization of which is activated here) of the proximate input. The function involved will determine both what features of the input have a bearing on the belief output and what bearing they have, i.e. how the content of the belief is determined by those features.’
p. 362 - ‘Again, consider inference to the best explanation. If I am trying to explain the presence of a pool of water on the floor of my basement, the relevant inputs will include not only beliefs about what the state of affairs is currently in my basement but also a list of possible causes of the water’s being there, and considerations that bear on the likelihood of each of these causes having been operative.’
Habits are belief forming…
p. 363 - We are interesrted in the operative function; the function that produces the particular belief.
p. 364 - ‘What the epistemic status of the particular belief depends on, according to reliabilism, is the conditional reliability of the operative mechanism, i.e., the extent to which the function realized by that mechanism would yeild true belief outputs from true inputs in a suitable spread of cases, plus the epistemic status of the input. Or, to put it in terms of processes, the crucial issue is the conditional reliability of the process of going from input to belief output in accordance with that function, along with the epistemic status of the inputs.’
Belief forming processes are embodied by a mechanism with an input-output process; ‘The function defines the epistemologically relevant type, and we can forget about the rest [such as 'processes that take place on Wednesday, et al].’
We work out the belief forming process based upon the output; the particular propostional content.
p. 365 - Assumes that there will only ever be one process per belief.
Makes the point that just because we can generate, after the fact, that a lot of different processes could have made us believe ‘that p’ that this does not mean that the belief was multiply realisable at the time.
Is a psychological realist (and claims you have to be for Reliabilism).
p. 366 - Multiple functions might work together to generate a belief.
p. 367 - ‘Why haven’t philosophers seen this, even philosophers like Goldman who realize that a belief-forming process is, essentially, the operation of a realized input-output function? Again, I suspect that the reluctance to be this pyschologically realisr plays a majot role here. If one simply talks of “processes,” each of which belongs to indefinitely many classes or types of processes, one is not making such strong psychological assumptions. But, as I said above, if one is not prepared to be that realist about the pyusche, one should quit trying to be a reliabilist in epistemology.’
p. 369 - ‘I have suggested that we think of belief formation in a pyschologically realistic way, as involving an input-output mechanism (habit) that yeilds belief outputs as a certain function of relevant features of inputs. If we do so, we can escape from the dilemma of what choice to make, for epistemic asessment as to reliability, of the type to which a particular belief-forming process belongs. That choice is settled for us by the identity of the function involved in the belief formation in question, for that function is something that possesses a built in generality. In other words, a particular belief formation is the activation of a general mechanism (habit) that operates in accordance with a certain function. The mechanism (habit) is the pyschological realization of that function. On this basis we can say what it is for a belief to be reliably formed.
I. A belief is reliably formed if and only if it was formed by the activation of a reliable belief-forming habit.
And:
II. A belief-forming habit is reliable if and nly if it would yeild a high proportion of true beliefs in a sufficiently large and varied run of exercises in situation of the sorts we typically encounter.’
