There are three conceptions of autonomy that Manson and O'Neill discuss: Kantian ``principled autonomy'', ``individual autonomy'', and ``rational autonomy''. In the following section, I outline the arguments they make in terms of informed consent against each of these, and discuss their dismissal of using any of these types of autonomy as the justification for practical informed consent procedures. Manson and O'Neill then dismiss autonomy altogether as a justification for informed consent, based on the dismissal of these three main types of informed consent.
Kantian ``principled autonomy'' is dismissed directly by Manson and O'Neill. To explain this simply, for Kant, autonomy is about self-imposition of the objective universal moral law, which is morally required of us because of our ability to practically reason. The ability to reason means that we can understand our freedom to make decisions, which Kant says must therefore be guided by a law that we will to follow, and which must be devoid of contingent influences. This is the (first) categorical imperative, that we must act only upon maxims that we would will to be universal law [Kant, 1785]. Imposing this upon oneself is, to Kant, autonomy. However, according to Manson and O'Neill, this isn't appropriate for use in medical ethics because the use of autonomy in a wider philosophical theory is not something that can be practically applied, or ``operationalised'' using informed consent procedures. I agree that this form of autonomy is not particularly useful as a basis for a theory of informed consent (particularly in IT), as its abstractness under-determines what is needed for its institutionalisation. Also, the emphasis on personal derivation of autonomy, through rational beings reflecting on the categorical imperative, then giving themselves the moral law qua rational agent, does not fit in here at all in any way that is useful, because we wish to set out guidelines for dealing with the general public, rather than allowing each person to establish their own moral law for these situations. People also do not generally behave purely in a Kantian fashion even if they wish to, since they are not always purely rational because they cannot remove their emotions from decision making. Ultimately the issue here for an information technology context is that there is a lack of fit between the general and demanding Kantian view and computer users making quick consent decisions in everyday IT settings.
The next type of autonomy, ``individual autonomy'', is a theory based on the principle that autonomy is a property of individuals: the independence of individuals (also mentioned by Appelbaum et al. (1987), above). This is something that informed consent principles can be applied to and protect, as individual choice is a part of a consent procedure, thus individual independence can be affected. However, Manson and O'Neill argue that protection of individual choice is also inappropriate as a justification for practical applications of informed consent, since there are many standardisations in areas like public health policy where individual choice cannot apply1.14, so to use it as a general justification for informed consent would be detrimental. Although Manson and O'Neill's theory below allows for individual choices to be protected to a certain degree, individual autonomy as a justification for informed consent is what is in fact being discounted, since this sort of autonomy requires people to be able to choose to opt out of public standardisations (such as public health, etc.). Some, they say, may wish to interpret the idea of individual autonomy as ``mere, sheer choice'' instead, calling for all choices to be protected, however uninformed. This strong libertarian view is unsuitable because informed consent procedures should be required to limit decisions to those who are informed, rather than accepting any decision.
The third type of autonomy, ``rational autonomy'', is initially problematic due to the number of conceptions of reason. Rational autonomy requires reflective, deliberative evaluation and being informed, or endorsing first-order desires by second-order volitions [Frankfurt, 1971]. First-order desires are the immediate desires (such as wanting to buy an MP3 player, or to eat sushi for lunch). When we desire a first-order desire to be effective in action, the second-order volition is what is identified with a free will. That is, in a sense, this is a reflective agent, distinct from the agent that might be governed by the ``mere, sheer choice'' of individual autonomy, who ``informs'' herself about the nature of her action through second-order volitions. However, similarly to the problem with principled autonomy, the difficulty of implementing informed consent based on these principles is high, due to the overwhelming need for education and cognitive, reflective decisions, which is difficult to implement across a wide base of people. Another issue is that with only this sort of reasoning taking place in the formation of the decision, things such as relief of suffering and beneficence would be dismissed as being only one of many lower-order desires to be weighted in a rational decision. Also, informed consent is a process that aims to protect actual choices by people, many of which, claim Manson and O'Neill, are not necessarily based on rational thought processes.
After evaluating these common understandings of autonomy1.15 as they apply to bioethics, Manson and O'Neill reject these as a basis for understanding informed consent. They point out that these justifications still use modern informed consent theory as their basis, but with a blind eye turned to the inadequacies of current bioethical standards. As we saw in the first section of this chapter, this is indeed the case, with autonomy being used in both of the more modern accounts of informed consent [Council of Europe, 1997,National Health and Medical Research Council et al., 2007]. Using autonomy as a justification, as is currently done, is an unattainable goal, which would require much stricter assessment of things like incompetence (rendering many currently competent individuals incompetent) and education (much higher standards would be needed). The other problem of using this sort of justification is that there are serious issues with using data previously collected in studies for a later study similar in intent but with a slightly different purpose; even the most minor of which would require re-consent by those involved, many of whom it may well be impossible to contact. Instead, they say, a new approach needs to be found that is ``both feasible and justifiable'' (p. 25).
I, too, agree that the concepts of autonomy outlined above seem problematic for use as the focus of consent. As we saw in the Faden and Beauchamp model, autonomy (of a rational form, although they do not use that terminology) forms a basic justification for the theory, and the arguments presented above are compelling in pointing to the underlying issues with using autonomy as the justification for informed consent in information technology. Furthermore, they imply that simply aiming for an autonomous action1.16 is a good enough end to reach, and informed consent naturally follows from that. However, as the practice shows in Chapter 2, there is a great difference between theoretical aims and practical applications of informed consent theory, and although the goal of sufficient autonomy seems to align with common sense, it is not a good idea in information technology to base the entire informed consent process on this vague and lofty (and ultimately impractical) goal, because ultimately it needs to be practical and, to a large extent, mechanisable.