Monday, March 20, 2006
Basil Mitchell on agency, telikenesis and detecting design
In his work on the design inference mathematician and philosopher William A. Dembski claims to have formalized (at least one of) the intuitive design detection apparatus of humanity. My confidence in the truth of this claim, and the claim that specified complexity is a reliable criterion of design detection, is boulstered by the fact that I keep coming accross academics who write about design detection and who employ what are more or less clearly pre-theoretic versions of Dembski's 'specified complexity' criterion. For example, arch Darwinist Richard Dawkins uses specified complexity as a design detector in Climing Mount Improbable, although he studiously ignores its implications.
My most recent discovery along these lines is a passage from Basil Mitchell's The Justification of Religious Belief (Macmillan, 1973). Mitchell was Nolloth Professor of the Philosophy of the Christian Religion and Fellow of Oriel College, Oxford prior to Richard Swinburne. In the course of defending the coherence of talking about incorporeal agency Mitchell has this to say on the subject of telikenesis (the alleged power to alter events such as the fall of dice by simply 'willing'):
'Whether or not telikenesis actually occurs, it does not seem difficult to specify the conditions under which we should be prepared to admit its occurence. If the dice were to fall with a certain number upwards whenever a particular individual was asked to bring it about and not otherwise, we should conclude that he had the power to cause physical changes without bodily movement. Bodily movement on the part of the agent is normally a reliable guide as to whether an occurence is an action or not, and, if so, whose; but we could, in principle, settle both questions without recourse to this criterion, if the other indications were clear enough. What are these? A combination of the following: (i) The unlikelihood of the event's occurrence apart from the intervention of some agent. (ii) The event's contributing to some purpose. (iii) The agreement of that purpose with the independently known character and purposes of the putative agent.' (p. 8.)
Note, first of all, that Mitchell is arguing that design can in principle be detected even if the design is not implimented by bodily agency.
Mitchell's design detection criterion has more parts than Dembski's, but then it attempts to do more, because it attempts to provide a criterion whereby we can detect not only that 'an occurrence is an action' but also 'whose' action it is. Mitchell's criterion for detecting intelligent design per se appears to be the same as Dembski's (albeit in a less outworked form).
Mitchell says that whether an occurence such as the falling of dice is an action can be answered positively if two conditions are met - and those conditions are sufficient complexity ('The unlikelihood of the event's occurrence apart from the intervention of some agent') combined with an independent specification ('specify the conditions under which we should be prepared to admit its occurence'/'The event's contributing to some purpose'). Knowledge concerning 'The agreement of that purpose with the independently known character and purposes of the putative agent', while helpful in pinning a designed event on a specific agent, is clearly not necessary for Mitchell's design inference per se.
Notice that this means that a design inference of the type advocated by Dembski does not of itself implicate any particular agent. Suppose our paranormal investigators set up some rigorous scientific experiments into telikenesis (would critics of ID condemn such experiments as anti-scientific in principle?) and the dice do indeed 'fall with a certain number upwards whenever a particular individual was asked to bring it about and not otherwise'. Suppose the specified complexity of this result exceeded Dembski's universal probability bound (something Mitchell doesn't bother calculating): while we should conclude that the best explanation for this result is intelligent design, we could not implicate our test subject on the basis of Dembski's design filter. Our test subject might have telikenetic powers, but the result we detected might have been caused by any agent. To settle on attributing the excersize of telikenetic powers in this instance to our test subject - rather than to God, or a god, or a ghost, or a demon, or an angel, or another human or alien with telikenetic powers in the next room who is trying to dupe our researchers into thinking that our subject has telikeneitc powers when he does not, etc. - our scientists must appeal to criteria beyond Dembski's design filter. Mitchell's 'agreement of that purpose with the independently known character and purposes of the putative agent' might be a good start here, but one imagines that Occam's razor should feature fairly heavily in such deliberations.
Basil Mitchell didn't clearly distinguish the criteria for infering design from the criteria for infering the responsibility of putative designers; he left his design detection criterion in a fairly pre-theoretic state (simply indicating the combination of low probability with a specification) sans the context of information theory and universal probability bounds deployed by Dembski, and, perhaps for these very reasons, Mitchell never made much of his criterion (and this in a book that discusses design arguments!). Nevertheless, it seems clear that Mitchell was thinking along the same lines as Dembski.
As far as one can tell, Dembski's thoughts on the matter are independent of Mitchell's. Now, the more scholars independently arrive at the same answer to a problem, the more confident we tend to be about the truth of their solution. Hence, discovering the use of specified complexity as a way to justify the inference to design as an aside in a work by a respected philosopher like Mitchell tends to add to our confidence that Dembski's answer to the question of design detection is correct.
My most recent discovery along these lines is a passage from Basil Mitchell's The Justification of Religious Belief (Macmillan, 1973). Mitchell was Nolloth Professor of the Philosophy of the Christian Religion and Fellow of Oriel College, Oxford prior to Richard Swinburne. In the course of defending the coherence of talking about incorporeal agency Mitchell has this to say on the subject of telikenesis (the alleged power to alter events such as the fall of dice by simply 'willing'):
'Whether or not telikenesis actually occurs, it does not seem difficult to specify the conditions under which we should be prepared to admit its occurence. If the dice were to fall with a certain number upwards whenever a particular individual was asked to bring it about and not otherwise, we should conclude that he had the power to cause physical changes without bodily movement. Bodily movement on the part of the agent is normally a reliable guide as to whether an occurence is an action or not, and, if so, whose; but we could, in principle, settle both questions without recourse to this criterion, if the other indications were clear enough. What are these? A combination of the following: (i) The unlikelihood of the event's occurrence apart from the intervention of some agent. (ii) The event's contributing to some purpose. (iii) The agreement of that purpose with the independently known character and purposes of the putative agent.' (p. 8.)
Note, first of all, that Mitchell is arguing that design can in principle be detected even if the design is not implimented by bodily agency.
Mitchell's design detection criterion has more parts than Dembski's, but then it attempts to do more, because it attempts to provide a criterion whereby we can detect not only that 'an occurrence is an action' but also 'whose' action it is. Mitchell's criterion for detecting intelligent design per se appears to be the same as Dembski's (albeit in a less outworked form).
Mitchell says that whether an occurence such as the falling of dice is an action can be answered positively if two conditions are met - and those conditions are sufficient complexity ('The unlikelihood of the event's occurrence apart from the intervention of some agent') combined with an independent specification ('specify the conditions under which we should be prepared to admit its occurence'/'The event's contributing to some purpose'). Knowledge concerning 'The agreement of that purpose with the independently known character and purposes of the putative agent', while helpful in pinning a designed event on a specific agent, is clearly not necessary for Mitchell's design inference per se.
Notice that this means that a design inference of the type advocated by Dembski does not of itself implicate any particular agent. Suppose our paranormal investigators set up some rigorous scientific experiments into telikenesis (would critics of ID condemn such experiments as anti-scientific in principle?) and the dice do indeed 'fall with a certain number upwards whenever a particular individual was asked to bring it about and not otherwise'. Suppose the specified complexity of this result exceeded Dembski's universal probability bound (something Mitchell doesn't bother calculating): while we should conclude that the best explanation for this result is intelligent design, we could not implicate our test subject on the basis of Dembski's design filter. Our test subject might have telikenetic powers, but the result we detected might have been caused by any agent. To settle on attributing the excersize of telikenetic powers in this instance to our test subject - rather than to God, or a god, or a ghost, or a demon, or an angel, or another human or alien with telikenetic powers in the next room who is trying to dupe our researchers into thinking that our subject has telikeneitc powers when he does not, etc. - our scientists must appeal to criteria beyond Dembski's design filter. Mitchell's 'agreement of that purpose with the independently known character and purposes of the putative agent' might be a good start here, but one imagines that Occam's razor should feature fairly heavily in such deliberations.
Basil Mitchell didn't clearly distinguish the criteria for infering design from the criteria for infering the responsibility of putative designers; he left his design detection criterion in a fairly pre-theoretic state (simply indicating the combination of low probability with a specification) sans the context of information theory and universal probability bounds deployed by Dembski, and, perhaps for these very reasons, Mitchell never made much of his criterion (and this in a book that discusses design arguments!). Nevertheless, it seems clear that Mitchell was thinking along the same lines as Dembski.
As far as one can tell, Dembski's thoughts on the matter are independent of Mitchell's. Now, the more scholars independently arrive at the same answer to a problem, the more confident we tend to be about the truth of their solution. Hence, discovering the use of specified complexity as a way to justify the inference to design as an aside in a work by a respected philosopher like Mitchell tends to add to our confidence that Dembski's answer to the question of design detection is correct.
